Support NeoGAF

[Karim]

Even on Pro, it's 60 with TAA, PSSR adds the extra overhead that is optional if you want a locked 60.

On SX it caps at 40 and it hits that cap 100% of the time.

And all that with RT.

That's better than 95% of games released in the last few years.

this boy's crazy.

Sure sure what about Alan Wake 2 in the 60 fps mode? It is new their engine favour the series X hardware? I bet whatever you want if they unlock the fps you will see a similar performance gap but sure I know you like to celebrate such stuff as a phenomenal event

 

[Bernkastel]

The Series X is over £200 cheaper than the PS5 Pro in the UK. Galaxy Black over £100 cheaper. And it is competitive in performance in many games. Xbox delivered good value this generation which was sadly undermined by tariffs.

But the only place where people buy Xbox is in the US.

 

[Darsxx82]

You mean the unlocked quality mode? If it was enough stable you could bet your balls there will be locked toggle to 60 fps...what I see here are the typical advantages of XSX hardware at 60 fps seen many times with the UE5...

It's not a UE5 game; the game in performance mode always runs well above 60fps and only 2-5fps slower than PRo with the same settings; quality mode on XSX never drops below 40fps with RT, which suggests a framerate very close to that of PS5PRo in that mode, too. But hey! there is no need to highlight anything because Karim says it !

 

[Karim]

It's not a UE5 game; the game in performance mode always runs well above 60fps and only 2-5fps slower than PRo with the same settings; quality mode on XSX never drops below 40fps with RT, which suggests a framerate very close to that of PS5PRo in that mode, too. But hey!!

You missed my point. This engine showed always better performance on XSX and I seen similar gap with UE5 at 60 fps, what exactly is shocking or surprising now? We already knew from awhile some engine perform better at 60 fps though most of the times is an unstable 60 fps so it's almost useless such advantage locking the 60 fps. And from what I remind the 60 fps mode was quite stable on ps5 base also. I mean it's fascinating to learn but practically doesn't change so much in the narrative of the hardware gap between the 2 console so I don't understand what you seen of beasty in XSX this time around

 

[adamsapple]

Sure sure what about Alan Wake 2 in the 60 fps mode? It is new their engine favour the series X hardware? I bet whatever you want if they unlock the fps you will see a similar performance gap but sure I know you like to celebrate such stuff as a phenomenal event

You missed my point. This engine showed always better performance on XSX

The irony of this sentence is that when Control first got its PS5/Series update, the engine actually ran better on PS5.

Beyond this, Series X exhibits some stutter not seen on PS5 - regardless of it being set to graphics or performance modes. It crops up with the arrival of UI elements on screen and in standard traversal, and can be distracting.

 

[Holammer]

- Update hits all console versions and PC.
*snip*

Is PC getting an update? DLSS framegen would be welcome, and a way to limit frame. You feel a noticeable controller judder when the game regularly jumps between 180-230fps.

 

[Zathalus]

This is important to mention... since many people stand still during gameplay

Static or slow movement is common in most cutscenes, and even in gameplay you don't always frantically move the character or camera around. Besides, the main issue with PSSR, namely the noise, is still present even in movement.

DF is once again right here, PSSR is an improvement over TAA but has a few flaws that still occur.

 

[MisterXDTV]

But the only place where people buy Xbox is in the US.

Used to buy maybe

Not anymore

 

[Panajev2001a]

The jaggies and the disocclusion artifacts.
PSSR looks much better than Remedy's TAA.

Yeah, but if you stare very closely with RT reflections there is some fizzling if you do not move the camera at all like you would do in real gameplay… </sarcasm>

Jokes aside, it is a massive step up… and in a few months we get a big update for it too (FSR4 update).

 

[winjer]

Jokes aside, it is a massive step up… and in a few months we get a big update for it too (FSR4 update).

I don't think that PSSR will just get the FSR4 update. It will probably also include Redstone, with ray-reconstruction, neural radiance cache and the ML frame generation.
And just recently, on an interview with the channel Ancient Gameplays, AMD lead marketing director said that Redstone would be coming "very soon".

 

[kevboard]

What was missing in your text about PSSR noise was

"especially in statistical moments."

This is important to mention... since many people stand still during gameplay

have you been in the graphics quality thread? because it sure as hell seems like a lot of people in there do nothing but stand still and stare at static shots. especially the UE5 Lumen defenders, as Lumen completely breaks in most games the moment anything moves too quickly

 

[skyfinch]

2025 is definitely the year of Xbox. First, listening to gamers about gamepass and adding more value. Now this. And there's still over 2 months left this year.

 

[viveks86]

I don't think that PSSR will just get the FSR4 update. It will probably also include Redstone, with ray-reconstruction, neural radiance cache and the ML frame generation. And just recently, on an interview with the channel Ancient Gameplays, AMD lead marketing director said that Redstone would be coming "very soon".

Would be awesome if true. But I just don't know if there is sufficient headroom for additional ML workloads on pro. If a "lightweight CNN" is already taking an extra ~2 ms, would they be able to do all that within the frame budget?

 

[Neo_game]

Good update for the Pro especially in gfx mode and PSSR. In Performance not sure people can tell difference between 65 and 75fps

 

[Radical_3d]

I don't think that PSSR will just get the FSR4 update. It will probably also include Redstone, with ray-reconstruction, neural radiance cache and the ML frame generation.

I'm the biggest Pro defender around here and this reads like condensed hopium to me. I'd love it to be true, tho.

 

[Three]

So in unlocked performance mode with same settings ps5pro is 22% faster than ps5 and 4% faster than xsx - 4 years after premiere for 750$, Mark Cerny brilliance at its peak ;d

I mean sure if you ignore the fact that the RT mode is 40fps on XSX/PS5 vs 60fps on PS5 Pro. A 50% difference. But yes in a mode at near 90fps the difference is going to be limited by other things like you said.

 

[winjer]

I'm the biggest Pro defender around here and this reads like condensed hopium to me. I'd love it to be true, tho.

I don't have a PS5 Pro, nor any other console, so I have no horse in this race.
But I do have a Radeon 9070, so I will be using Redstone as soon as it launches.

Considering the Amethyst partnership, there is a good chance some of these techs will show up on the Pro.
FSR4 Int8 is already confirmed. FSR3 frame generation was already used on a few games, so it's possible that FSR ML Frame Generation might also make a showing on the Pro.
The neural radiance cache could be huge to improve performance with RT games.

 

[Radical_3d]

I don't have a PS5 Pro, nor any other console, so I have no horse in this race.
But I do have a Radeon 9070, so I will be using Redstone as soon as it launches.

Considering the Amethyst partnership, there is a good chance some of these techs will show up on the Pro.
FSR4 Int8 is already confirmed. FSR3 frame generation was already used on a few games, so it's possible that FSR ML Frame Generation might also make a showing on the Pro.
The neural radiance cache could be huge to improve performance with RT games.

The 9070 is PS6 tier hardware. Cerny said that the Pro had to find memory in some hidden part of the architecture to make the ML work. If they try to implement something that isn't upscaling would be worsening that feature, maybe? I don't know but the best way to not be disappointed is to expect nothing.

 

[winjer]

The 9070 is PS6 tier hardware. Cerny said that the Pro had to find memory in some hidden part of the architecture to make the ML work. If they try to implement something that isn't upscaling would be worsening that feature, maybe? I don't know but the best way to not be disappointed is to expect nothing.

True, but it's also a GPU that runs games at 100+ fps at 1440p.
A console like the Pro, for most games targets 30 or 60 fps.

 

[Darsxx82]

I mean sure if you ignore the fact that the RT mode is 40fps on XSX/PS5 vs 60fps on PS5 Pro. A 50% difference. But yes in a mode at near 90fps the difference is going to be limited by other things like you said.

This is because Quality mode is locked at 40fps. That 50% difference isn't real.
First, because XSX shows it has room to grow, and not a single fps is detected below 40fps. And second, PRo is far from 60fps and drops to 40s during demanding moments. In other words, what's incomprehensible is why it's locked at 40fps on base consoles (especially on XSX).

There's nothing like a 50% difference in performance, and the real % difference is probably closer to what's seen in performance mode.

 

[Three]

?? D Darsxx82 why have you quoted my post with literally nothing?

 

[Darsxx82]

?? D Darsxx82 why have you quoted my post with literally nothing?

Sorry, I don't know what happened...but I accidentally hit "reply" when I started typing.... Things with the mobile text bar

It's already fixed.

 

[MisterXDTV]

I'm the biggest Pro defender around here and this reads like condensed hopium to me. I'd love it to be true, tho.

Well Cerny already said the algorythm is 100% the same as FSR 4 full-fat

About the other parts of "Redstone" I have no idea whether they are part of Project Amethyst or not

For sure the ML/Ray Tracing part of the PS5 Pro GPU is RDNA 4. It's 9070 level

Other parts of the GPU are limited to RDNA 2.x to keep seamless compatibility with base PS5

 

[Three]

This is because Quality mode is locked at 40fps . That 50% difference isn't real.
First, because XSX shows it has room to grow, and not a single fps is detected below 40fps.

They limit it to 40fps for a reason and yes it does drop below 40fps on Series X. I'm not sure where you're getting that idea from

And second, PRo is far from 60fps and drops to 40s during demanding moments. In other words, what's incomprehensible is why it's locked at 40fps on base consoles (especially on XSX).

There's nothing like a 50% difference in performance, and the real difference is probably closer to what's seen in performance mode.

No it's not, it's up to 50% in real terms. You can try to make some imaginary reason why it's capped at 40 but at the end of the day XSX is capped to 40fps when PS5 Pro hits 60. You can even see this performance difference is real when XSX goes below the cap.

 

[Darsxx82]

They limit it to 40fps for a reason and yes it does drop below 40fps on Series X. I'm not sure where you're getting that idea from

No it's not, it's up to 50% in real terms. You can try to make some imaginary reason why it's capped at 40 but at the end of the day XSX is capped to 40fps when PS5 Pro hits 60. You can even see this performance difference is real when XSX goes below the cap.

This is specifically The famous "Corridor of Doom" while in the general game and mayor/big action-packed scenes, XSX doesn't drop below 40fps while PRo stays at 40s....

Then during gameplay, PRo constantly hovers at 50-60fps. That is, it's far from a stable 60fps. To think that XSX, with those fps figures (and with the data from performance mode) isn't comfortably running above 40fps is, I'm sorry, nonsense.

Why haven't they enabled an "60fps" Quality mode on XSX? I don't know. What's clear is that it doesn't make sense to not do it also the VRR extension on XS allowed it perfectly.

PS: There are other games that have enabled (in a discreet and unofficial manner) it on XSX. Wukong, for example, recently due balanced mode 1440p with Lumen running 50-55fps average on 60hz VRR mode.

 

[Three]

This is specifically The famous "Doom Corridor," while in the general game and big action-packed scenes, XSX doesn't drop below 40fps while PRo stays at 40s....

Then during gameplay, PRo constantly hovers at 50-60fps. That is, it's far from a stable 60fps. To think that XSX, with those fps figures (and with the data from performance mode) isn't comfortably running above 40fps is, I'm sorry, nonsense.

Why haven't they enabled an "60fps" Quality mode on XSX? I don't know. What's clear is that it doesn't make sense to not do it also the VRR extension on XS allowed it perfectly.

PS: There are other games that have enabled it on XSX. Wukong, for example, recently.

You do know why it's capped at 40fps. You just don't want to accept the reason. You claimed "not a single fps is below 40fps" I showed you this isn't true. I showed that without the cap coming into effect there is a 43% performance advantage to the PS5 Pro. You were claiming it's the same as performance mode at 90fps ie 5%. That's what's nonsense. What you think/believe it could be running at above that cap is irrelevant too, at the end of the day the PS5 pro is showing 50% more fps through large sections of the game. That is a real performance difference in practical terms.

 

[adamsapple]

You do know why it's capped at 40fps. You just don't want to accept the reason. You claimed "not a single fps is below 40fps" I showed you this isn't true. I showed that without the cap coming into effect there is a 43% performance advantage to the PS5 Pro. You were claiming it's the same as performance mode at 90fps ie 5%. That's what's nonsense. What you think/believe it could be running at above that cap is irrelevant too, at the end of the day the PS5 pro is showing 50% more fps through large sections of the game. That is a real performance difference in practical terms.

ehhhhhh

The fact that they're capped at different limits doesn't really give us any idea of a realistic difference between PS5/SX and PS5 Pro in the Quality mode.

That'd be like saying that BF6 on SX is giving a nearly 100% improvement in performance because some of the scripted scenes are capped to 60 on PS5/Pro but unlocked to 120 on Xbox.

 

[Pedro Motta]

Remedy's engine favours DX12 based machines, nothing new here.

 

[Three]

ehhhhhh

The fact that they're capped at different limits doesn't really give us any idea of a realistic difference between PS5/SX and PS5 Pro in the Quality mode.

So you think D Darsxx82 claiming its a 5% difference is Ok when I showed that without the cap in effect (at 37fps) the PS5 Pro is running at 43% higher fps?

Yes you don't know the exact % difference in sections when the cap is in effect but claiming its 5% and not knowing why they chose a 40fps cap throughout the entire game is nonsense. Come on. It's because it's hovering close to 40 through most of it just like the PS5 base but PS5 drops a little more to go below.

He also claimed it "isn't real" but this is the empirical real difference in the game. The theoretical difference he is alluding to is irrelevant just as it is in some of those BF scripted cutscenes or whatever you're referring to.

 

[MisterXDTV]

He also claimed it "isn't real" but this is the empirical real difference in the game. The theoretical difference he is alluding to is irrelevant just as it is in some of those BF scripted cutscenes or whatever you're referring to.

They are desperate at this point...

Let them cry alone in their rooms

 

[adamsapple]

So you think D Darsxx82 claiming its a 5% difference is Ok when I showed that without the cap in effect (at 37fps) the PS5 Pro is running at 43% higher fps?

Yes you don't know the exact % difference in sections when the cap is in effect but claiming 5% and not knowing why they chose a 40fps cap throughout the entire game is nonsense. Come on. It's because it's hovering close to 40 through most of it just like the PS5 base but PS5 drops a little more to go below.

I'm not sure if D Darsxx82 was referring to this, but DF have done some unlocked FPS testing on Control before as well, without the 40fps cap.

The game's photo mode unlocked the FPS and was something of a benchmark and in most test cases, at least the SX version, would show at or around the ~50fps or higher mark. With that one noted 'Corridor of doom' area noted to drop as low as 33fps back then, now that area seems to run around 38~fps so it's not that hard to infer that the rest of the areas would have seen optimization improvements as well.

This is the most stressed area in their comparison back then. Now this area seems to run much better per the latest patch.

Had they let the SX version run unlocked to 60, it would have stayed within the VRR window almost all of the time and would have been a great way to play.

Why they didn't ? Who can say, it may be because they wanted to keep parity between PS5 and SX or a different reason.

He also claimed it "isn't real" but this is the empirical real difference in the game. The theoretical difference he is alluding to is irrelevant just as it is in some of those BF scripted cutscenes or whatever you're referring to.

Sure, just like saying there's a 'a real performance difference in practical terms' when the comparison samples aren't capped to the same metrics.

 

[Darsxx82]

You do know why it's capped at 40fps. You just don't want to accept the reason.

Or It's rather you who wants to convince yourself that there was no room for a "60fps" VRR mode on XSX even with the data and VRR characteristics because ahem...

There are games that are officially locked at 40fps, but thanks to the way VRR works on the XS, we've been able to enjoy unlocked 60fps VRR modes. And, as is understandable and logical, games that run at 40fps without any drop actually show a frame rate that's significantly higher. You have the recent example of Wukong, which in balanced mode 1440p (1080p on PS5) sits at 50-55fps on 60hz, and is perfect with VRR

You claimed "not a single fps is below 40fps" I showed you this isn't true. I showed that without the cap coming into effect there is a 43% performance advantage to the PS5 Pro.

No, you've only shown the one instance where that happens, and in an area where nothing happens and which is famous for that reason.
Could you post screenshots where Pro drops to 40s and where XSX doesn't drop below 40fps... Or do you mean those moments (which are generally the most demanding) aren't valid?

You were claiming it's the same as performance mode at 90fps ie 5%. That's what's nonsense.

No, there's no need to distort what was said. I said that the FPS data in Performance mode is a good basis for intuiting what would happen in Quality mode as well, and that it's much more plausible that the margins are closer to what we saw in that mode than the 50% you're claiming as a real difference.

What you think/believe it could be running at above that cap is irrelevant too,

It's 2025, and I think everyone knows that a game capped at 40fps without a drop in the most demanding moments is running comfortably above that figure the vast majority of the time, and that a game locked at 60fps that drops to the mid-40s is not far off.
So no, it's not irrelevant, it's the reason why we're not seeing the real difference, which is certainly not 50% that you defend.

at the end of the day the PS5 pro is showing 50% more fps through large sections of the game. That is a real performance difference in practical terms.

Yes, "in practical terms," not "in real terms."

 

[Darsxx82]

They are desperate at this point...

Let them cry alone in their rooms

 

[Donald mac Ronald]

If you take a turd.

Polish that turd, shine it up real good. Give it the old spit shine.

It's still a turd that's so shiny you can see yourself in it's ray traced reflection.

 

[adamsapple]

If you take a turd.

Polish that turd, shine it up real good. Give it the old spit shine.

It's still a turd that's so shiny you can see yourself in it's ray traced reflection.

Good thing then that we're talking about Control, an amazing game.

 

[rnlval]

Well Cerny already said the algorythm is 100% the same as FSR 4 full-fat

About the other parts of "Redstone" I have no idea whether they are part of Project Amethyst or not

For sure the ML/Ray Tracing part of the PS5 Pro GPU is RDNA 4. It's 9070 level

Other parts of the GPU are limited to RDNA 2.x to keep seamless compatibility with base PS5

Wrong, Digital Foundry placed PS5 Pro (60 CU scale RDNA 2/RDNA 3) as about on par with 32 CU scale RDNA 4.0 (with double rate TMUs per CU backing wave32 dual issue mode) RX 9060 XT 16GB. RX 9060 XT's dual-issue TFLOPS is useful when coupled with double-rate TMUs.


PlayStation 5 Pro vs AMD RX 9060 XT vs Nvidia RTX 5060 Ti!

RDNA 4.0's ML is on par with NVIDIA's ADA Lovelace i.e. 4:2 sparsity FP8 feature. PS5 Pro doesn't support sparsity and FP8.

Unlike PC's RDNA 4 RX 9060/9070 series, PS5 Pro does not support FP8 natively, hence the FSR4-based PSSL update is delayed into 2026.

 

[SKYF@ll]

Has this update fixed the ray tracing reflection bug on Xbox Series X?
Previously, the PS5 version looked a bit more detailed. (Look at the reflection on the floor)

 

[Three]

I'm not sure if D Darsxx82 was referring to this, but DF have done some unlocked FPS testing on Control before as well, without the 40fps cap.

The game's photo mode unlocked the FPS and was something of a benchmark and in most test cases, at least the SX version, would show at or around the ~50fps or higher mark. With that one noted 'Corridor of doom' area noted to drop as low as 33fps back then, now that area seems to run around 38~fps so it's not that hard to infer that the rest of the areas would have seen optimization improvements as well.

This is the most stressed area in their comparison back then. Now this area seems to run much better per the latest patch.

Had they let the SX version run unlocked to 60, it would have stayed within the VRR window almost all of the time and would have been a great way to play.

Why they didn't ? Who can say, it may be because they wanted to keep parity between PS5 and SX or a different reason.

DF were showing you the large differences not the sections that warranted the reason for the 30fps cap of the past, they even said there are good reasons for the 30fps cap in that past video. Why do you think that is? Because staring at a wall in photomode isn't the taxing parts. The latest updates brought a 40fps cap and DF themselves were asking for that based on improving the past performance.

No, you've only shown the one instance where that happens, and in an area where nothing happens and which is famous for that reason.

It's not the one instance it happens. I just showed you that one when you made the claim "there isn't a single fps below 40".

No, there's no need to distort what was said. I said that the FPS data in Performance mode is a good basis for intuiting what would happen in Quality mode as well, and that it's much more plausible that the margins are closer to what we saw in that mode than the 50% you're claiming as a real difference.

I don't need to distort what was said. This is exactly what you said:

There's nothing like a 50% difference in performance, and the real difference is probably closer to what's seen in performance mode.

The real difference in performance mode is ~5% when it is hitting near 90fps and being limited by things other than the GPU power.

explain why there is a 43% difference when the 40fps cap is not in effect and how that is closer to 5% rather than 50%. You forget that the PS5 Pro is capped at 60fps too so it too may be running higher there. What makes you claim this ridiculous idea that the difference is closer to 5% without evidence when there is evidence to the contrary when it falls below the cap?

 

[adamsapple]

DF were showing you the large differences not the sections that warranted the reason for the 30fps cap of the past, they even said there are good reasons for the 30fps cap in that past video. Why do you think that is? Because staring at a wall in photomode isn't the taxing parts. The latest updates brought a 40fps cap and DF themselves were asking for that based on improving the past performance.

If you go by this video, they say (on SX at least) the game stays locked to 40 outside of the corridor of doom area that you capped above.

Almost all games that target 60 often have dips from it too, that one area is the exception not the norm.

My point is had the game been unlocked to 60, not 40, it would have been a perfectly viable way to play with a VRR display.

As it is, I'm happy with the really good 40 that they have. It's a massive improvement over the 30 cap.

 

[omegasc]

My point is had the game been unlocked to 60, not 40, it would have been a perfectly viable way to play with a VRR display.

Yeah, capping to 40 does not make sense on Series X + VRR screen when it's basically always on target.

 

[Three]

If you go by this video, they say (on SX at least) the game stays locked to 40 outside of the corridor of doom area that you capped above.

It mostly keeps above 40, nobody is denying that but there are other sections where it drops below too. just because the DF video doesn't show it doesn't mean it doesn't exist.

I'm referring to the old 30fps unlock framerate in photomode you showed. They said they did it for framerate consistency (DF were just showing you the larger differences in averages in that older sponsored video outside of gameplay), if you look at the frametime graph it was jumping between 33ms (30fps) so you're seeing an average fps with poor framepacing in that photomode where it doesn't really matter if it is inconsistent and there isn't anything going on vs in game. Now look at the frametime graph for the new 40fps mode and it is consistent 25ms in game. That's a good thing and even DF were asking for modes like that. It made sense to go for a 40fps mode just as it did 30fps in the past patch because that's the consistent frametime they could guarantee. It's not hovering close to 57fps (17ms) (which would be what it would need to achieve to be close to a 5% difference vs 60fps). Otherwise they would have just done a 60fps mode. Do you believe it is mostly hovering at 57fps and they simply chose a 40fps cap? Then why would you agree with Darsxxx. Especially when you have evidence when the cap doesn't apply.

 

[S0ULZB0URNE]

ehhhhhh

The fact that they're capped at different limits doesn't really give us any idea of a realistic difference between PS5/SX and PS5 Pro in the Quality mode.

That'd be like saying that BF6 on SX is giving a nearly 100% improvement in performance because some of the scripted scenes are capped to 60 on PS5/Pro but unlocked to 120 on Xbox.

It's not unlocked in single player for PS5 Pro.

 

[adamsapple]

Has this update fixed the ray tracing reflection bug on Xbox Series X?
Previously, the PS5 version looked a bit more detailed. (Look at the reflection on the floor)

Yep, can confirm first hand it's been fixed now. Contrast difference from my direct capture aside.

 

[Fafalada]

RDNA 4.0's ML is on par with NVIDIA's ADA Lovelace i.e. 4:2 sparsity FP8 feature. PS5 Pro doesn't support sparsity and FP8.

Sparsity isn't a feature - it's simply not running compute on the 'empty' parts of the matrices.
The reason it wasn't a thing on older GPUs is because wide-parallel compute doesn't always have granular enough flow control. But as Mark explained it last year, PS5 Pro does.

But yes - converting floating to fixed point math does require some rework of an algorithm.

- PSSR toggle added to PS5 Pro with a cost of 2.2ms on rendering budget, Remedy mentions a drop in performance in patch notes

One caveat here - that's 2.2ms cost against no TAA. Relative to Remedy's TAA it's about 1.2ms (the number is validated throughout DFs capture samples), ie. TAA itself is about 1ms on the Pro in given settings.
Nothing particularly new (FSR also lands in that ballpark in the past), but just for extra info.

 

[rnlval]

Sparsity isn't a feature - it's simply not running compute on the 'empty' parts of the matrices.
The reason it wasn't a thing on older GPUs is because wide-parallel compute doesn't always have granular enough flow control. But as Mark explained it last year, PS5 Pro does.

But yes - converting floating to fixed point math does require some rework of an algorithm.


One caveat here - that's 2.2ms cost against no TAA. Relative to Remedy's TAA it's about 1.2ms (the number is validated throughout DFs capture samples), ie. TAA itself is about 1ms on the Pro in given settings.
Nothing particularly new (FSR also lands in that ballpark in the past), but just for extra info.

Sparsity is a feature. Ask NVIDIA. Hardware sparsity support has been included with Ampere. FP8 was included with ADA LoveLace.

 

[kevboard]

It's not unlocked in single player for PS5 Pro.

that's the point. you can't judge the performance of the SX version if we don't have a true unlocked option.
a fully unlocked quality mode on SX would probably feel way better than the current 40fps lock in 90% of the scenarios in the game. and LFC is already built into the system.

it only really seems to drop below 40fps in the "corridor of doom" which has almost the whole screen filled with both transparencies, transparent RT reflections and opaque RT reflections. so shitloads of overdraw across a large area of the screen.
even if you only reach 50fps on average, that will feel noticeably smoother.

 

[Fafalada]

Sparsity is a feature.

PS3 can do ML with sparsity...
PS2... can compute ML with sparsity.

Out of all the SIMD compute accelerators in consoles released for last 25 years, more of them do sparsity before NVidia even 'invented' ML than after.

 

[S0ULZB0URNE]

that's the point. you can't judge the performance of the SX version if we don't have a true unlocked option.
a fully unlocked quality mode on SX would probably feel way better than the current 40fps lock in 90% of the scenarios in the game. and LFC is already built into the system.

it only really seems to drop below 40fps in the "corridor of doom" which has almost the whole screen filled with both transparencies, transparent RT reflections and opaque RT reflections. so shitloads of overdraw across a large area of the screen.
even if you only reach 50fps on average, that will feel noticeably smoother.

I was agreeing I worded it wrong.

 

[Three]

a fully unlocked quality mode on SX would probably feel way better than the current 40fps lock in 90% of the scenarios in the game. and LFC is already built into the system.

it only really seems to drop below 40fps in the "corridor of doom" which has almost the whole screen filled with both transparencies, transparent RT reflections and opaque RT reflections. so shitloads of overdraw across a large area of the screen.
even if you only reach 50fps on average, that will feel noticeably smoother.

Nothing beats the smoothness of the 40fps they opted for. The consistent frametimes they've opted for with their engine is the smoothest you're going to get. VRR isn't smoother than a consistent frametime. VRR is a method to try and make inconsistent frametime more palatable and LFC is an absolute last resort. They both still come with drawbacks vs a nice consistent locked framerate.

 

[kevboard]

Nothing beats the smoothness of the 40fps they opted for. The consistent frametimes they've opted for with their engine is the smoothest you're going to get. VRR isn't smoother than a consistent frametime. VRR is a method to try and make inconsistent frametime more palatable and LFC is an absolute last resort. They both still come with drawbacks vs a nice consistent locked framerate.

that's just not true. I played multiple games now at unlocked framerates that hover in the 40~60 fps range, and all of them felt and looked better unlocked, than locked... to be fair, there is no way to lock Elden Ring, but comparatively, unlocked Elden Ring feels smoother than Control at 40fps.
another game I played unlocked is Spider-Man 2. I played the quality (usually 40fps) mode unlocked... and it just is better overall than locking it to 40fps

and it's not like it is a smooth 40fps on PS5 for example, so what's the excuse there? why not lock it to 30fps only on PS5? especially since, if it doesn't have built in LFC, you will have stutters there.