• 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.

DF: Cyberpunk 2077 Next-Gen Patch: The Digital Foundry Verdict

Some of the drops are like 10 fps though, So weird. Does this mean PS5 has more CPU power available? So weird. The rez differences are so minuscule too, it has to be because CDreject are no good.

RT mode drops below 30 FPS on Series X, so weird how RT shadows and higher quality SSR are so taxing on the cpu. The more you know.
Actually we know that RT is actually taxing the CPU
 
Last edited:

Tripolygon

Banned
It's possible that the way the PS5 "distribute" the power between the GPU and CPU is more efficient, and the CPU on XsX is simply throttling under heavy load such as some scenes in this game.
Consoles do not throttle like how PCs do. They would sooner shut down than throttle performance. There could a whole host of reasons for the performance issues but throttling is certainly not one of them.
 

SlimySnake

Flashless at the Golden Globes
It's possible that the way the PS5 "distribute" the power between the GPU and CPU is more efficient, and the CPU on XsX is simply throttling under heavy load such as some scenes in this game.
The XSX has fixed tflops and a fixed CPU clock. They have enough budget to keep the CPU operating at peak speeds. You are thinking of the PS5 which downclocks the CPU and GPU in some rare instances.

On traditional consoles like the PS4, Pro, and XSX, if the CPU or GPU is getting too hot, the console would simply shut down to prevent any damage. You can test this by removing the fan in the PS4 pro.

What's happening here is a bottleneck due to some kind of compromises made in the design of the console. It's like their split ram architecture or the GPU being unable to feed its CUs properly. Something Cerny brought up in his Road to PS5. There is no 52 CU RDNA 2.0 card with the same architecture as the XSX. Their 13 tflops card uses 40 CUs and boost the clocks to get to 13 tflops.
 

VFXVeteran

Banned
VFXVeteran VFXVeteran has shared his IMDB and YouTube channel with us before, it's not a secret, but if he wants I can delete those posts.
Someone said he was cosplaying as a vetted VFX guy but that's not the case.
Yes, I have shared it and yes I was vetted by the powers that be that controls these forums. They know my work and they know the contacts I have in the gaming industry.

I don't understand why my credentials keep coming up from the Sony guys when I speak about 3D graphics and the limitations in hardware (which is clearly shown by review sites). It's ok for them to disagree with what I'm saying but they are taking it to personal levels when they have to try denouncing my credentials. I shouldn't feel like I have to prove myself to anyone here everytime there is something that someone dislikes about my comments. Especially when they have 0 experience in developing 3D graphics (realtime or otherwise).
 
Last edited:

sendit

Member
Michael Jackson Popcorn GIF by Naughty Dog
 

SlimySnake

Flashless at the Golden Globes
Elden Rings is also having framerate issues on the series x. Neither the PS5 or the XSX is a locked 60 fps, but early comparisons before the day one patch show the xsx is around 48 while the ps5 is around 55.



We will see what happens when the day one patch drops but this is an interesting point of comparison before optimizations are made.

P.S We dont have resolution numbers so the XSX could have a higher resolution.
 

Mr Moose

Member
Elden Rings is also having framerate issues on the series x. Neither the PS5 or the XSX is a locked 60 fps, but early comparisons before the day one patch show the xsx is around 48 while the ps5 is around 55.



We will see what happens when the day one patch drops but this is an interesting point of comparison before optimizations are made.

P.S We dont have resolution numbers so the XSX could have a higher resolution.

Not sure if I'd trust an old version of trdrop.
 

Lysandros

Member
Elden Ring beta was actually slightly higher resolution on PS5 along with (clearly) higher performance according to DF analysis. Also XSX lacked motion blur.

 
Last edited:

ethomaz

Banned
Elden Rings is also having framerate issues on the series x. Neither the PS5 or the XSX is a locked 60 fps, but early comparisons before the day one patch show the xsx is around 48 while the ps5 is around 55.



We will see what happens when the day one patch drops but this is an interesting point of comparison before optimizations are made.

P.S We dont have resolution numbers so the XSX could have a higher resolution.

Most if not all lately releases has framerate better on PS5.
It is basically a trend already.
Maybe the tools are not there yet.
 
Last edited:

ChiefDada

Gold Member
I don't understand why my credentials keep coming up from the Sony guys when I speak about 3D graphics and the limitations in hardware (which is clearly shown by review sites). It's ok for them to disagree with what I'm saying but they are taking it to personal levels when they have to try denouncing my credentials. I shouldn't feel like I have to prove myself to anyone here everytime there is something that someone dislikes about my comments. Especially when they have 0 experience in developing 3D graphics (realtime or otherwise).

I've NEVER questioned your credentials. You often say things that are objectively false. The basis of our counter claims aren't derived from our experience or lack thereof, but from developers who actually work on the hardware. I wish you would knock it off with the "Sony fans" rhetoric, you use it as an attempt to minimize the verifiable facts brought forth by those who disagree with you. You have the potential to be a valuable resource for so many people here with your tech knowledge, but instead you use it as an authoritative badge to misinform others.
 

ChiefDada

Gold Member
Elden Rings is also having framerate issues on the series x. Neither the PS5 or the XSX is a locked 60 fps, but early comparisons before the day one patch show the xsx is around 48 while the ps5 is around 55.



We will see what happens when the day one patch drops but this is an interesting point of comparison before optimizations are made.

P.S We dont have resolution numbers so the XSX could have a higher resolution.


I would've thought the embargo placed on digital foundry as it relates to day 1 patch would've also been applied to other reviewers as well.

Either way, I'm assuming day 1 patch solves this so not a big issue.
 

PaintTinJr

Member
The XSX has fixed tflops and a fixed CPU clock. They have enough budget to keep the CPU operating at peak speeds. You are thinking of the PS5 which downclocks the CPU and GPU in some rare instances.
....
The TF are fixed at a utilisation that doesn't exceed the thermal or power budgets that would get the system to shutdown, so semantically the same as throttling the clock, as it is an A = B x C triangle equation. Just like V = I x R or P = V x I. By stating in the way you have, you make it sound like it can do 12TF with full workloads continuously, when the power supply and/or cooling system couldn't handle that at all as Cerny explained in the post Road to PS5 interview with DF - when Richard asked a question of that very nature.
 

assurdum

Banned
The XSX has fixed tflops and a fixed CPU clock. They have enough budget to keep the CPU operating at peak speeds. You are thinking of the PS5 which downclocks the CPU and GPU in some rare instances.

On traditional consoles like the PS4, Pro, and XSX, if the CPU or GPU is getting too hot, the console would simply shut down to prevent any damage. You can test this by removing the fan in the PS4 pro.

What's happening here is a bottleneck due to some kind of compromises made in the design of the console. It's like their split ram architecture or the GPU being unable to feed its CUs properly. Something Cerny brought up in his Road to PS5. There is no 52 CU RDNA 2.0 card with the same architecture as the XSX. Their 13 tflops card uses 40 CUs and boost the clocks to get to 13 tflops.
Oh Jesus not again this bullshit of fixed clock has better performance. It's not true at all.
 
Last edited:

SlimySnake

Flashless at the Golden Globes
You think because CPU has fixed clock XSX can't have inefficiency compared ps5 because it's "variable". That's what I get it from your post
Not even close.

Read this part again.
What's happening here is a bottleneck due to some kind of compromises made in the design of the console. It's like their split ram architecture or the GPU being unable to feed its CUs properly. Something Cerny brought up in his Road to PS5. There is no 52 CU RDNA 2.0 card with the same architecture as the XSX. Their 13 tflops card uses 40 CUs and boost the clocks to get to 13 tflops.

The fixed clocks bit is in reply to the poster's assertion that the XSX is getting throttled under heavy load. The only time the XSX or the PS4 or any other console not named the PS5 is getting throttled is if they get too hot at which point the console itself shuts down completely.

It's possible that the way the PS5 "distribute" the power between the GPU and CPU is more efficient, and the CPU on XsX is simply throttling under heavy load such as some scenes in this game.
 
Last edited:

SlimySnake

Flashless at the Golden Globes
The TF are fixed at a utilisation that doesn't exceed the thermal or power budgets that would get the system to shutdown, so semantically the same as throttling the clock, as it is an A = B x C triangle equation. Just like V = I x R or P = V x I. By stating in the way you have, you make it sound like it can do 12TF with full workloads continuously, when the power supply and/or cooling system couldn't handle that at all as Cerny explained in the post Road to PS5 interview with DF - when Richard asked a question of that very nature.
Not at all. I am simply pointing out that throttling of CPUs or GPUs in a console with fixed clocks is simply not something that can happen unless it shuts down which almost never happens.

I am making zero claims about how the XSX always runs at its highest clocks 100% Or the PS5 does not. Just ruling out speculation that the XSX's performance issues are due to throttling of CPU. It's like saying the base PS4 dropped frames in Cyberpunk because its 1.6 ghz CPU was being throttled somehow. Nah, that simply does not happen in fixed clocks consoles. They pick the CPU and GPU max clocks and leave themselves enough headroom in the cooling to go over the pre-determined max temps like Cerny said just in case some weird game heats up the GPU more than your typical game. IIRC, he used an example of a racing game vs an action adventure game. Doesnt mean, the racing game wasnt operating at peak clocks.
 
Last edited:

Bogroll

Likes moldy games
What's happening here is a bottleneck due to some kind of compromises made in the design of the console. It's like their split ram architecture or the GPU being unable to feed its CUs properly.
So what's your theory when SX outperforms the PS5 in the indoor shootout ?
 

SlimySnake

Flashless at the Golden Globes
So what's your theory when SX outperforms the PS5 in the indoor shootout ?
Timestamp?

They looked the same to me in this video. The only drop I noticed on the PS5 was in NX Gamer's video where shattering a glass door caused drops to 46 fps on both consoles but thats an aberration than norm.
 

Bogroll

Likes moldy games
Timestamp?

They looked the same to me in this video. The only drop I noticed on the PS5 was in NX Gamer's video where shattering a glass door caused drops to 46 fps on both consoles but thats an aberration than norm.
Yes NXGamer. I don't recall SX dropping as low as the PS5 in that clip. Who's to say it don't happen else where. They just seem to focus on the known trouble spots at the start of the game and driving around where the SX will drop 2 frames when there doesn't seem much happening. Also there is a point when the buildings fade out on the PS5 and not on SX. Which I pointed out in a thread but conveniently got ignored.
But it still outperformed (from what I remember)the PS5 in that NXgamer video so whats going on there.

I just watched again drops to 46 PS5 52 SX and he also says "Sometimes in firefights even ever so slightly better" That implies more than one to me. Maybe not every time but now and then but obviously I'm guessing.
 
Last edited:

SlimySnake

Flashless at the Golden Globes
Yes NXGamer. I don't recall SX dropping as low as the PS5 in that clip. Who's to say it don't happen else where. They just seem to focus on the known trouble spots at the start of the game and driving around where the SX will drop 2 frames when there doesn't seem much happening. Also there is a point when the buildings fade out on the PS5 and not on SX. Which I pointed out in a thread but conveniently got ignored.
But it still outperformed (from what I remember)the PS5 in that NXgamer video so whats going on there.
Just watched it again. It is 46 vs 52 for a fraction of a second. It's just a stutter if anything not a sustained tflops. It's what I was talking about earlier on the last page, those rare drops dont mean much. They could be due to alpha effects, maybe the way the glass shattered on the PS5, you see more particles on screen vs the xsx. The PS5 has consistently done better in other games when alpha particles fill up the screen so again, just an abberation.

The NX Gamer tests show pretty much a locked 60 fps indoors on the PS5 and XSX which likely means they are both running way above 60 fps. Cant really glean much from that.

The test DF ran is basically running through the game world or driving through it which is likely hitting the streaming system hard and we know that is PS5's main strength.
 
Last edited:

Bogroll

Likes moldy games
Just watched it again. It is 46 vs 52 for a fraction of a second. It's just a stutter if anything not a sustained tflops. It's what I was talking about earlier on the last page, those rare drops dont mean much. They could be due to alpha effects, maybe the way the glass shattered on the PS5, you see more particles on screen vs the xsx. The PS5 has consistently done better in other games when alpha particles fill up the screen so again, just an abberation.

The NX Gamer tests show pretty much a locked 60 fps indoors on the PS5 and XSX which likely means they are both running way above 60 fps. Cant really glean much from that.

The test DF ran is basically running through the game world or driving through it which is likely hitting the streaming system hard and we know that is PS5's main strength.
Its still down for a few seconds. Question being why is SX outperforming there presumably at a higher resolution.
 

Riky

$MSFT
Its still down for a few seconds. Question being why is SX outperforming there presumably at a higher resolution.

Lots of effects which have big gains on Tier 2 VRS, the drops on Series X outdoors are in heavy NPC areas so are CPU related. If they were GPU related the DRS scaler would kick in and lower the resolution, but it doesn't so you can rule out the GPU and any bizarre theories about ram banks and bandwith.
 

Bogroll

Likes moldy games
Lots of effects which have big gains on Tier 2 VRS, the drops on Series X outdoors are in heavy NPC areas so are CPU related. If they were GPU related the DRS scaler would kick in and lower the resolution, but it doesn't so you can rule out the GPU and any bizarre theories about ram banks and bandwith.
But surely its the same as the PS5 cpu. Well at least close enough to not make that much of a difference. The test down death alley or whatever it was called wasn't thorough enough for my liking. Not saying they were being dodgy or anything but he just skimped on it imo.
 
Lots of effects which have big gains on Tier 2 VRS, the drops on Series X outdoors are in heavy NPC areas so are CPU related. If they were GPU related the DRS scaler would kick in and lower the resolution, but it doesn't so you can rule out the GPU and any bizarre theories about ram banks and bandwith.
I think using cyberpunk to gauge for bottlenecks is quite silly. This is a game that runs much better in slow HDD mode regardless of hard drive type or speed. The drops for xsx are actually quite similar to PC with fast hard drive mode selected. Maybe that's the difference
 

Bogroll

Likes moldy games
I think using cyberpunk to gauge for bottlenecks is quite silly. This is a game that runs much better in slow HDD mode regardless of hard drive type or speed. The drops for xsx are actually quite similar to PC with fast hard drive mode selected. Maybe that's the difference
Thats mad. Am I reading that correctly ? It runs better off a HDD than SSD ?
 
Thats mad. Am I reading that correctly ? It runs better off a HDD than SSD ?
No it runs better in slow HDD mode. Its an option. I have a sn850 and I get zero hitches in slow HDD mode but decent drops in fast HDD mode. My kid with a 1060 and slow nvme got about 20 more fps just switching to slow HDD mode haha.
He was getting drops to 20 with a 40 avg. Now it's closer to locked 60

I'm saying now but I mean months ago when we all played it.
 
Last edited:

PaintTinJr

Member
Not at all. I am simply pointing out that throttling of CPUs or GPUs in a console with fixed clocks is simply not something that can happen unless it shuts down which almost never happens.

....
The console's OS work scheduler or the game developers will have a function that conservatively slows the rate of work being scheduled for the XsX silicon when lower threshold thermal and power limits are reached, reactively doing the same as downclocking, so the terminology or exact wording seems like a word game when the end results is the silicon reactively being able to do less work when thermal/power limits are reached, so semantically no different IMHO.

What is different is the PS5's paradigm shift of a workload analyser that knows the amount of power available is fixed and that the clock to do the work can be set to consume all that energy as a simple act of balancing the a = b x c equation that the developers know and are pre-emptively designing against.
 
Last edited:

adamsapple

Or is it just one of Phil's balls in my throat?
I think this game is far from final on all platforms. I would bet that we see further performance improvements on all platforms including pc over the coming months.
Pretty much.

A years worth of patches have brought it to a state we can call it acceptable and enjoyable on consoles.

They'll continue working on it with the DLC and expansions.
 

Lysandros

Member
The console's OS work scheduler or the game developers will have a function that conservatively slows the rate of work being scheduled for the XsX silicon when lower threshold thermal and power limits are reached, reactively doing the same as downclocking, so the terminology or exact wording seems like a word game when the end results is the silicon reactively being able to do less work when thermal/power limits are reached, so semantically no different IMHO.

What is different is the PS5's paradigm shift of a workload analyser that knows the amount of power available is fixed and that the clock to do the work can be set to consume all that energy as a simple act of balancing the a = b x c equation that the developers know and are pre-emptively designing against.
Exquisite posts, hope these will broaden the horizons a bit.. 👍
 

Razvedka

Banned
But surely its the same as the PS5 cpu. Well at least close enough to not make that much of a difference. The test down death alley or whatever it was called wasn't thorough enough for my liking. Not saying they were being dodgy or anything but he just skimped on it imo.
PS5 has a net CPU advantage from what I recall. Between offloading more work and cache scrubbing. I also wouldn't be surprised if the PS SDK is just better/more efficient then what MS can offer.

It's been a while since I was involved in reading about the hardware side of these next gen consoles though.
 
Last edited:

Synless

Member
Yeah and it's also Cyberpunk, a game that needed a years worth of patches to get in an acceptable state in the first place :messenger_tears_of_joy:

And i'll also post this since some of y'all only care about PS5 vs SX. This is the actual improvement. Look at the frame graph on the 1.23 patch picture and then compare it with 1.50. This is the most important improvement over the patch, not the difference of ~100p in SX's favor in Performance mode or the fluctuating FPS improvement here and there on PS5.



BQUfHyI.png
It being cyberpunk has nothing to do with the comment, the point I see that poster making is many games run better on PS5 and given the parameters the poster your responding to mentions, the question is why? Almost all games are on par with ps5 with negligible wins at best or losing to it. I’m more interested in that answer than any of the console war shit, it shouldn’t happen based on specs alone, but here we are.
 

adamsapple

Or is it just one of Phil's balls in my throat?
It being cyberpunk has nothing to do with the comment, the point I see that poster making is many games run better on PS5 and given the parameters the poster your responding to mentions, the question is why? Almost all games are on par with ps5 with negligible wins at best or losing to it. I’m more interested in that answer than any of the console war shit, it shouldn’t happen based on specs alone, but here we are.

Well, the question you're asking has no reasonable answer that won't devolve into the console war shit you're not interested in.

The easiest answer is that these two consoles are the closest we've seen in a while, the TF differential itself is less than Pro and One X, and they both have varying strengths and weaknesses, one of the PS5's that we've known from a while before the consoles even came out is that it is easier to develop for. The Xbox GDK are also a limiting factor and believe it or not "THE TOOLS" were also a factor, not just a meme.

Referring to this classic example, Control on Xbox launched with some stutter and pauses, but over DF's coverage the issues fixed themselves and the game actually became more preferable on Xbox over PS5. And this happened *without* the game being updated. Only system updates since launch.



"Just by the Xbox Series X getting updated over time, we'll see games getting performance boost. The issue wasn't the game itself but the Xbox development stack.....
I've heard from other developers that through Xbox updates they've seen improvements in performance on ray tracing.... "

This feeds directly into the "PS5 is easier to develop for out of the gate" point. Won't be surprised if this is a PS3/360 scenario where the 360 was easier to develop for out of the gate and had advantages in multiplatform games in the launch years, but they became equal over time and better on PS3 in some cases. The true strengths will always be the exclusives, preferably first party ones which fully utilize the hardware.


-

That being said, now this will turn into more console war bullshit again.
 
Last edited:

Lysandros

Member
PS5 has a net CPU advantage from what I recall. Between offloading more work and cache scrubbing. I also wouldn't be surprised if the PS SDK is just better/more efficient then what MS can offer.
I think a combination of factors, be it lower interconnect latencies (as stated by Hargett), less cycle requirements due to I/O hardware, (possibly) more sound processing taken care of by Tempest and lower overhead API give PS5 CPU an edge more often than not. In the end i see both machines as being neck to neck, be it GPU, CPU or RAM/bandwidth overall, with I/O hardware & SSD being the exeption.
 
Last edited:

Hayabusa83

Banned
I have an unopened Series X version, now that I am seeing all these comparisons I am questioning if I made the right decision on which console to get it for.
 

adamsapple

Or is it just one of Phil's balls in my throat?
I have an unopened Series X version, now that I am seeing all these comparisons I am questioning if I made the right decision on which console to get it for.

Like the post above said, if you have a VRR or Freesync display connected to your Xbox, it's the easy choice. It also has the added benefit of a slightly higher average resolution as per DF.

But if you don't have a VRR display and can't stand the somewhat extra frame drops, swap for the PS5 version.
 

azertydu91

Hard to Kill
Like the post above said, if you have a VRR or Freesync display connected to your Xbox, it's the easy choice. It also has the added benefit of a slightly higher average resolution as per DF.

But if you don't have a VRR display and can't stand the somewhat extra frame drops, swap for the PS5 version.
Do you have VRR on your TV? Keep it, if not, exchange it for the PS5 version.
Hey stop being reasonable you two, this is gaf we war here!!!
So there's really not gonna be a green rat ™* or a blue dildo?

*green rats is a deposed mark legally possessed by kretos kretos any copy or unauthorized use will be followed by legal actions.
 

Hugare

Member
Well shiet. On my pc (5930k and 1080ti) I can only get 40ish most of the time, but it doesnt go below high 30s...
It's a damn good port, imo

My notebook with a RTX 2060 doesnt come near 60 fps at 1080p high settings + low RT, even with DLSS

And these consoles are running the game at 1440p locked 60

I would have enjoyed the inclusion of a "quality RT mode" tho, with all the RT features at 1080/30, since RT Lighting and Reflections really change this game
 
Top Bottom