Support NeoGAF

[sncvsrtoip]

Yep. Compute heavy engine like UE5 finally shows Xbox GPU advantage, quite consistently in recent releases.

it has 13% more fps in this scene over base ps5 ;d hard to compare to pro as it has higher pixels (up to 32%) with outside of selective scene perf is similar

In the scene adamsapple showed it's almost 22%:

Pro has a bit higher resolution but worse framerate.

and what is shocking with that ? xsx has ~25% faster gpu, pro all around doesnt have worse frame rate (in some scenes has more in slectevie few worse), generaly I see pattern that less talented studios more pc experienced deliver better perf on xbox as its api is closer to pc

 

[Bojji]

it has 13% more fps in this scene over base ps5 ;d hard to compare to pro as it has higher pixels (up to 32%) with outside of selective scene perf is similar

and what is shocking with that ? xsx has ~25% faster gpu, pro all around doesnt have worse frame rate (in some scenes has more in slectevie few worse), generaly I see pattern that less talented studios more pc experienced deliver better perf on xbox as its api is closer to pc

Less talented devs... yeah.

Pro in many games seems memory BW limited, same can be true for base PS5. Xbox has smaller GPU than Pro but almost the same memory BW.

 

[sncvsrtoip]

Less talented devs... yeah.

Well yeah, Im Pole and wish Bloober team all best but they are not (yet) in top tier league

 

[Bojji]

Well yeah, Im Pole and wish Bloober team all best but they are not (yet) in top tier league

But this is not just blooper, many games on UE5 perform like that (Mafia, MGS3). And UE5 is as current gen engine as it gets, it's very heavy on compute units. Older engines that rely more on faster clocks of fixed hardware units may perform better on PS5 (and they often do).

Xbox also gets less love from developers because it's a failed platform, why would they optimize for it when PS5 has ~3x more potential customers?

 

[SkyHighTrees]

But this is not just blooper, many games on UE5 perform like that (Mafia, MGS3). And UE5 is as current gen engine as it gets, it's very heavy on compute units. Older engines that rely more on faster clocks of fixed hardware units may perform better on PS5 (and they often do).

Xbox also gets less love from developers because it's a failed platform, why would they optimize for it when PS5 has ~3x more potential customers?

so you're saying that if they did get the tools, optimized for xbox as lead platform, harnessed the power of the cloud they could have a game running at 30fps on ps5 and 60fps on the xbox?

daaaaaaaaaaaaym

 

[Bojji]

so you're saying that if they did get the tools, optimized for xbox as lead platform, harnessed the power of the cloud they could have a game running at 30fps on ps5 and 60fps on the xbox?

daaaaaaaaaaaaym

I never said that, and you know it.

 

[sncvsrtoip]

But this is not just blooper, many games on UE5 perform like that (Mafia, MGS3). And UE5 is as current gen engine as it gets, it's very heavy on compute units. Older engines that rely more on faster clocks of fixed hardware units may perform better on PS5 (and they often do).

Xbox also gets less love from developers because it's a failed platform, why would they optimize for it when PS5 has ~3x more potential customers?

Thats the point they dont optimise they are just more experienced in pc dev. I cant find other explanation xsx has any fps advantages over ps5pro in any scene (one case could be like in Mgs delta they choose setting wronlgy with pssr etc that even base ps5 outperform pro, but is not a case in Cronos). Btw pro is not even officialy supported in Cronos. Its only backward compatibilty with brute performance advantage.

 

[Bojji]

Thats the point they dont optimise they are just more experienced in pc dev. I cant find other explanation xsx has any fps advantages over ps5pro in any scene (one case could be like in Mgs delta they choose setting wronlgy with pssr etc that even base ps5 outperform pro, but is not a case in Cronos).

This game doesn't have specific Pro "enhanced" version for some reason. It's running PS5 code on bigger GPU (it even has the same 2560x1440 upper resolution).

 

[adamsapple]

This game doesn't have specific Pro "enhanced" version for some reason. It's running PS5 code on bigger GPU (it even has the same 2560x1440 upper resolution).

Let's be honest, it's UE5, it's Bloober, their version of 'Pro Enhancements' would have been another bad implementation of PSSR like SH2 Remake before they yanked it off.

This is probably the best case scenario.

 

[Mr Moose]

But this is not just blooper, many games on UE5 perform like that (Mafia, MGS3). And UE5 is as current gen engine as it gets, it's very heavy on compute units. Older engines that rely more on faster clocks of fixed hardware units may perform better on PS5 (and they often do).

Xbox also gets less love from developers because it's a failed platform, why would they optimize for it when PS5 has ~3x more potential customers?

Virtuos did two of the recent games. They don't seem very good with the engine. Bloober aren't either.

 

[SlimySnake]

Yep. Compute heavy engine like UE5 finally shows Xbox GPU advantage, quite consistently in recent releases.

Nah. if it was compute then the PS5 Pro would benefit a lot more. they have 60 CUs compared to the 52 in XSX. Running at much higher clocks too.

what you are seeing is the 18% advantage finally manifesting in games. tbh, we saw it in some games earlier this gen. it just feels like its happening more often because we are literally getting a UE5 game every other week.

 

[kevboard]

Nah. if it was compute then the PS5 Pro would benefit a lot more. they have 60 CUs compared to the 52 in XSX. Running at much higher clocks too.

the pro probably suffers from the same old same old, devs setting the resolution boundaries too high.

same shit constantly happend on the Xbox One X. Sekiro had a lower framerate on One X than on PS4 Pro, because they doubled the resolution... sure in some cases that kinda worked on One X (RDR2 as an example), but often it didn't.

if the Pro ran at the exact same resolution as the Series X, it would have an FPS advantage, but it doesn't.

in that first pixel count the Pro is pushing 2,742,336 pixels, while the Series X only renders 2,073,600 pixels.
that's a 32% increase in pixel count

 

[adamsapple]

Nah. if it was compute then the PS5 Pro would benefit a lot more. they have 60 CUs compared to the 52 in XSX. Running at much higher clocks too.

what you are seeing is the 18% advantage finally manifesting in games. tbh, we saw it in some games earlier this gen. it just feels like its happening more often because we are literally getting a UE5 game every other week.

Whenever Rogue City is mentioned, it's always brought up that the Xbox version has a tarmac texture loading issue so it's not 'better'.

But I own both the XB and (base) PS5 versions and recently played through on the PS5 less than a month ago, that version runs like SHIT. The farther you get into the game, the more combat heavy scenarios, the frame rate drops to what feels like the teens. It only really reaches a smooth level in completely empty indoors. The SX version, by comparison, runs much better. PS5's narrower VRR window doesn't help ironing out the deficiencies either.

Yes, you'll see some users show a picture of how it drops to 24fps in a late game combat scenario, but you can imagine how the PS5 version runs there if it's 10~ fps behind just running around the city.

 

[SlimySnake]

the pro probably suffers from the same old same old, devs setting the resolution boundaries too high.

same shit constantly happend on the Xbox One X. Sekiro had a lower framerate on One X than on PS4 Pro, because they doubled the resolution... sure in some cases that kinda worked on One X (RDR2 as an example), but often it didn't.

if the Pro ran at the exact same resolution as the Series X, it would have an FPS advantage, but it doesn't.

ah ok. that makes more sense that. i thought they were the same resolution. but ps5 pro has a 46% increase in pixels compared to the ps5 which is kinda hilarious because its very close to the 45% average increase Cerny mentioned. i still cant believe he saw those 45% benchmarks for his 67% faster GPU and said to himself, yeah, we are good. No bottlenecks here. Awful engineering. Guy has clearly lost it.

 

[DenchDeckard]

You can talk all you want about what you want...

But when the amazing 12 TF beast is getting outsold big time by the fucking Switch 1 in 2025, it is pointeless

Nobody buys the console, nobody buys games on that console....

So in the end, the question is really...

Who gives a fuck about the 12 TF dead console???

You sound salty as foook.

 

[SlimySnake]

Whenever Rogue City is mentioned, it's always brought up that the Xbox version has a tarmac texture loading issue so it's not 'better'.

But I own both the XB and (base) PS5 versions and recently played through on the PS5 less than a month ago, that version runs like SHIT. The farther you get into the game, the more combat heavy scenarios, the frame rate drops to what feels like the teens. It only really reaches a smooth level in completely empty indoors. The SX version, by comparison, runs much better. PS5's narrower VRR window doesn't help ironing out the deficiencies either.

Yes, you'll see some users show a picture of how it drops to 24fps in a late game combat scenario, but you can imagine how the PS5 version runs there if it's 10~ fps behind just running around the city.

lots of bullshit in those early years. i feel you. im no xbox fan but im one of those tflops are tflops guys and i caught some strays back then too.

 

[adamsapple]

lots of bullshit in those early years. i feel you. im no xbox fan but im one of those tflops are tflops guys and i caught some strays back then too.

To be absolutely fair, there were some odd examples near the start of the gen that definitely didn't help Xboxes cause, the biggest one that comes to mind is AC Valhalla which ran up to 10 frames faster on the PS5 and also without any tearing, the Xbox version ran worse and had full screen tearing at launch in that game.

 

[Bojji]

Let's be honest, it's UE5, it's Bloober, their version of 'Pro Enhancements' would have been another bad implementation of PSSR like SH2 Remake before they yanked it off.

This is probably the best case scenario.

Yep, maybe they wanted no shit storm this time around. Better fps and resolution is still way more "enhanced" than that MGS3 remake, lol.

Virtuos did two of the recent games. They don't seem very good with the engine. Bloober aren't either.

According to Tim Sweeney no one is good with their engine.

Nah. if it was compute then the PS5 Pro would benefit a lot more. they have 60 CUs compared to the 52 in XSX. Running at much higher clocks too.

what you are seeing is the 18% advantage finally manifesting in games. tbh, we saw it in some games earlier this gen. it just feels like its happening more often because we are literally getting a UE5 game every other week.

Teraflops mostly cover compute power because you need clocks + number of shader cores. It doesn't include things like rops, memory bandwith etc.

I think Pro is not very well designed - Cerny probably had his hands tied when it comes to some aspects (cost).

 

[Radical_3d]

The performance boost in Pro is nice but I'm not having FSR on my screen anymore, thanks.

 

[kevboard]

The performance boost in Pro is nice but I'm not having FSR on my screen anymore, thanks.

it's using TSR!

 

[kazuyamishima]

FSR4 coming to PS5 Pro in 2026 can't come soon enough. But if devs have to patch things up like with PSSR I don't have too much hope.

 

[adamsapple]

The performance boost in Pro is nice but I'm not having FSR on my screen anymore, thanks.

Unreal's built-in TSR is pretty good in most cases.

Especially for being a software only solution.

 

[Radical_3d]

it's using TSR!

Well it looks like a Nutella toast for me when there is fire, explosions or fast movement. So it's still a hard pass. And I don't even have a very good sight. I can't see the differences in resolution of the 3 machines once reconstructed to 4K. And between PS5 and Pro there is even a 40 something percentage difference according to Tom in the video. But the pixel smears pierces through YouTube compression and my aging eyes like a hot white knife on butter.

Unreal's built-in TSR is pretty good in most cases.

Especially for being a software only solution.

See above.

 

[kevboard]

Well it looks like a Nutella toast for me when there is fire, explosions or fast movement. So it's still a hard pass. And I don't even have a very good sight. I can't see the differences in resolution of the 3 machines once reconstructed to 40K. And between PS5 and Pro there is even a 40 something difference according to Tom in the video. But the pixel smears pierces through YouTube compression and my aging eyes like a hot white knife on butter.

well, that's what happens when the GI, the AO, the transparencies, the shadows, the reflections and the volumetrics are all dithered and need to literally be smeared through a denoiser and TAA to look coherent.

combine that with a sub 1440p, and sometimes sub 1080p resolution, and what you get is the typical UE5 pixel mush.

not sure if PSSR would have helped here tbh.

 

[willothedog]

You can talk all you want about what you want...

But when the amazing 12 TF beast is getting outsold big time by the fucking Switch 1 in 2025, it is pointeless

Nobody buys the console, nobody buys games on that console....

So in the end, the question is really...

Who gives a fuck about the 12 TF dead console???

did someone drop you on your head as a child?

 

[Bojji]

Well it looks like a Nutella toast for me when there is fire, explosions or fast movement. So it's still a hard pass. And I don't even have a very good sight. I can't see the differences in resolution of the 3 machines once reconstructed to 4K. And between PS5 and Pro there is even a 40 something percentage difference according to Tom in the video. But the pixel smears pierces through YouTube compression and my aging eyes like a hot white knife on butter.

Even if they used PSSR - it would have issues for sure. I'm playing hell is us on Pro right now and game has PSSR/TSR switch, it looks BAD with PSSR (like most UE5 games).

So what is left: TSR and FSR3. TSR is the best choice.

 

[Karim]

But this is not just blooper, many games on UE5 perform like that (Mafia, MGS3). And UE5 is as current gen engine as it gets, it's very heavy on compute units. Older engines that rely more on faster clocks of fixed hardware units may perform better on PS5 (and they often do).

Xbox also gets less love from developers because it's a failed platform, why would they optimize for it when PS5 has ~3x more potential customers?

Not this bullshit again. It's no matter of older engine. The gap you see here it's exactly the difference of raw power 18/22% of advantage in favour of series X. There is nothing of new.

 

[Bojji]

Not this bullshit again.

Convince me that's is not the case.

 

[Radical_3d]

well, that's what happens when the GI, the AO, the transparencies, the shadows, the reflections and the volumetrics are all dithered and need to literally be smeared through a denoiser and TAA to look coherent.

combine that with a sub 1440p, and sometimes sub 1080p resolution, and what you get is the typical UE5 pixel mush.

not sure if PSSR would have helped here tbh.

It would. Maybe is not perfect but is a generational leap over software solutions.

Even if they used PSSR - it would have issues for sure. I'm playing hell is us on Pro right now and game has PSSR/TSR switch, it looks BAD with PSSR (like most UE5 games).

So what is left are: TSR and FSR3. TSR is the best choice.

It has issues but is better. And hell is us is not this game. We don't know how it would fair, although judging from their previous game and implementation probably like superimpose a Locoroco level over the whole image.

 

[Karim]

Convince me that's is not the case.

Most of the games show always such difference from awhile. Sometimes it's resolution sometimes fps sometimes both. PS5 occasionally beat series X, series X more often but not all the engines are the same, not necessarily are older the engine which perform better on ps5.

 

[Bojji]

It would. Maybe is not perfect but is a generational leap over software solutions.

It has issues but is better. And hell is us is not this game. We don't know how it would fair, although judging from their previous game and implementation probably like superimpose a Locoroco level over the whole image.

Most UE5 devs don't touch PSSR.

Out of all UE5 games I only know about Wukong looking good with it. Hell is Us is bad (I don't know why they implemented it) and SH2 was shit.

 

[Radical_3d]

Convince me that's is not the case.

Ah… yea… the old aging architecture of the Pro… God knows why a 16TF machine dips with fire into the 40s, but probably that YouTube guy that exposes UE5 developers would do a video soon.

Most UE5 devs don't touch PSSR.

Out of all UE5 games I only know about Wukong looking good with it. Hell is Us is bad (I don't know why they implemented it) and SH2 was shit.

Doesn't that Ninja Theory demo use PSSR?

 

[Bojji]

Ah… yea… the old aging architecture of the Pro… God knows why a 16TF machine dips with fire into the 40s, but probably that YouTube guy that exposes UE5 developers would do a video soon.

My guess is memory bandwith. Fire is bandwidth heavy.

 

[Karim]

Ah… yea… the old aging architecture of the Pro… God knows why a 16TF machine dips with fire into the 40s, but probably that YouTube guy that exposes UE5 developers would do a video soon.

Because triple buffering if I can say.

 

[Bojji]

Because triple buffering if I can say.

Triple buffering don't drop performance, it adds more latency.

Triple buffering aims to balance performance and visual quality, particularly with V-Sync enabled, by using a third buffer to prevent the GPU from becoming idle and to reduce screen tearing, while a potential drawback is an increase in input lag compared to double buffering. It works by allowing the GPU to render frames independently of the display's refresh rate, improving frame delivery and frame rates, but this comes at the cost of potentially adding one frame of latency.

 

[Karim]

My guess is memory bandwith. Fire is bandwidth heavy.

I'm confuse. Doesn't ps5 pro and Series X has a similar amount of CUs and bandwidth? Why it's not a problem on Series X?

 

[Bojji]

I'm confuse. Doesn't ps5 pro and Series X has a similar amount of CUs and bandwidth? Why it's not a problem of Series X?

Lower resolution maybe? Xbox also has separate BW for CPU stuff so 560GB/s goes all into GPU while Pro has to share it with CPU.

 

[Radical_3d]

My guess is memory bandwith. Fire is bandwidth heavy.

So, your theory is that a 576GBs machine is bandwidth constrained over a 10GBs only 560GBs machine.

 

[Karim]

Lower resolution maybe? Xbox also has separate BW for CPU stuff so 560GB/s goes all into GPU while Pro has to share it with CPU.

But this game it's not pro enhanced it uses the same range of DRS of series X.

 

[Karim]

So, your theory is that a 576GBs machine is bandwidth constrained over a 10GBs only 560GBs machine.

Yeah I mean ps5 pro has bandwidth bottleneck issues but series X not when they are more close in CUs and bandwith setting than Series X and base ps5. I really don't follow this logic.

 

[adamsapple]

But this game it's not pro enhanced it uses the same range of DRS of series X.

In all the DF comparisons, Pro is always running higher DRS*. It probably has some tweaked values to run on Pro even if it isn't a bespoke "Pro" enhanced game.

* Not the biggest of differences tho.

 

[Radical_3d]

Lower resolution maybe? Xbox also has separate BW for CPU stuff so 560GB/s goes all into GPU while Pro has to share it with CPU.

This is not how it works. May someone correct me (Kepler or someone like that) but SX is not a 990GB/s machine. Is a 560GB/s tops machine with some modules being slower. The interface has its number of lines and if you occupy them, like any north bridge, then some of the bandwidth is not disposable. Also, if that would be the case the Pro should also add the DDR5 memory bandwidth that uses for the OS and such.

 

[Karim]

In all the DF comparisons, Pro is always running higher DRS*. It probably has some tweaked values to run on Pro even if it isn't a bespoke "Pro" enhanced game.

* Not the biggest of differences tho.

It's just backward.

 

[Bojji]

So, your theory is that a 576GBs machine is bandwidth constrained over a 10GBs only 560GBs machine.

But this game it's not pro enhanced it uses the same range of DRS of series X.

Even when game is dropping to the lowest it's still 8.5% higher resolution on Pro for some reason, maybe what DF suspects is wrong and game is aware that it's running on Pro.

And in gameplay it has higher res most of the time:

DRS also takes some time to react, that might be the reason why Pro drops lower when there is fire - it starts with higher resolution when it happens.

 

[Karim]

Even when game is dropping to the lowest it's still 8.5% higher resolution on Pro for some reason, maybe what DF suspects is wrong and game is aware that it's running on Pro.

And in gameplay it has higher res most of the time:

DRS also takes some time to react, that might be the reason why Pro drops lower when there is fire - it starts with higher resolution when it happens.

So DRS not works properly on ps5 pro probably because it just optimized for the ps5 base version. I guess.

 

[Radical_3d]

DRS also takes some time to react, that might be the reason why Pro drops lower when there is fire - it starts with higher resolution when it happens.

 

[adamsapple]

DRS also takes some time to react, that might be the reason why Pro drops lower when there is fire - it starts with higher resolution when it happens.

Yeah.

Pro is both the highest average performant version, and also the version which has the single biggest reported drop/stretch.

That DRS gets stressed in that fiery barrier explosion scene.

 

[SKYF@ll]

Nah. if it was compute then the PS5 Pro would benefit a lot more. they have 60 CUs compared to the 52 in XSX. Running at much higher clocks too.

what you are seeing is the 18% advantage finally manifesting in games. tbh, we saw it in some games earlier this gen. it just feels like its happening more often because we are literally getting a UE5 game every other week.

Robocop is not a good benchmark comparison.
I've checked it in several situations and it seems like several settings are different between PS5 and XSX. *Latest version
DF "PS5: It's also rendering a more complex road surface which could be making reflections especially expensive to compute."

RoboCop Rogue City (Quality Mode) PS5:1440p vs. XSX:1080p *the asphalt textures in the downtown area don't display correctly on Series X

RoboCop Rogue City (Performance Mode) PS5:1080p vs. XSX:1080p *Compare signs and car windows.

 

[Radical_3d]

That DRS gets stressed in that fiery barrier explosion scene.

Alpha effects where at its peak with the PS2 bandwidth. Since then the bandwidth never caught the raster incremental and now that flimsy effect that the PS2 would just do is a stress test for a thing that in 2001 we would have say that renders real life.

They should have that in mind. I'm all out for future proofing your games but maybe test your shaders in current hardware a little? Anyways, this effect catches poor Pro with 1200p and therefore the face planking.

Anyways I'm 170+ hours into Elden Ring SotET, a 20 hours horror adventure months before RE9 would be refreshing. Fix that TSR shit, Bloober, and you have a sale.

 

[kevboard]

RoboCop Rogue City (Performance Mode) PS5:1080p vs. XSX:1080p *Compare signs and car windows.

it looks like the Series X has chromatic aberration enabled, while the PS5 doesn't.

notice how the "blur" increases the further you look left. while the further right you look in that image, the more similar they look.

that game is generally a technical mess however, especially if you aren't on PC, as the only way to get this to have decent image quality os using DLSS