Support NeoGAF

[Bonsaibäumchen]

Are you comparing textures only?

Yes as stated.
I've read somewhere textures are the biggest chunk thats constantly loaded on hdds and about 25% of all data thats usually loaded.
However this was last gen. I have no idea how things will be on next-gen. Hard to tell.

Either way this example was all in favor of xsx anyways. My guess is PS5's speed advantage will be bigger in real life environments if its neccesary. Because the 60% compression of bcpqck will only effect 25% of the overall data thats beeing loaded.
The rest will use zlip which has about 25% less compression then kraken.

Xsx: 100GB Data = 25GB Textures @ 60% Compression + 75GB Other Data @ 22.5% Compression = 10GB compressed Textures + 58.125 GB other Data = 68.125GB @ 2.4GB/s = 28.39seconds

PS5: 100GB @ 30% compression = 70GB @ 5.5GB/s = 12.73 seconds

Just another take at this. Theres no way we can say which scenario games will use in the future. Either way PS5's SSD will always be faster. Even in the worst case its still 24% faster then xsx.
In this case here its even about 55% faster.

 

[rnlval]

1. That .5 difference was ~40%, the current 1.8 is ~15%, if math serves me right

2. More FUD. This has been explained tons of times by people that have way more insight on the tech behind the clock speed.

3. If you're talking about current game design I wholeheartedly agree. But if you start leveraging the SSD for new ways of designing your games, you will absolutely see a difference (in first party games)

4. Nobody skimped on GPU. The faster Xbox knuckleheads understand that both consoles were designed differently, the faster you'll realize that both of them will perform similarly in third party games.

1. The difference magnitude remains similar.
GPU's FLOPS
X1X vs PS4 Pro = 1.8 TFLOPS GCN
XSX vs PS5 = 1.872‬ TFLOPS RDNA

PS4 Pro to PS5 TFLOPS increase = 6.075‬ TFLOPS
X1X to XSX TFLOPS increase = 6.147‬ TFLOPS

GPU's memory bandwidthfor render targets
X1X vs PS4 Pro = 102 GB/s
XSX vs PS5 = 112 GB/s

PS4 Pro to PS5 memory bandwidth increase = 224 GB/s
X1X to XSX memory bandwidth increase = 234 GB/s

The percentage argument hides the actual magnitude difference.

The next-generation consoles shifted to a higher level with similar gains when compared X1X and PS4 Pro respectively.

 

[Three]

As I pointed out, texture streaming is not seldom used. It is common. Just google “[engine name] texture streaming”. The streaming is the same as what you’re talking about - moving mip levels or tiles in and out of memory.



1) Texture sampling hardware was extended to allow writing back what was sampled. Previously texture sampling hardware could not do this. This was clearly stated in the video I linked.

2) Not sure what “not show on new hardware” means. Do you mean why it might not be on the PS5? Because it’s a brand new feature for AMD and it’s not out of the question that it isn’t in their RDNA implementation (And other stated reasons).



Yes, in order to cover terrain with a massive texture:

id Tech 4 - Wikipedia

en.wikipedia.org

”The MegaTexture technology tackled this issue by introducing a means to create expansive outdoor scenes. By painting a single massive texture (32,768×32,768 pixels, though it has been extended to larger dimensions in recent versions of the MegaTexture technology) covering the entire polygon map and highly detailed terrain, the desired effects can be achieved.”

The aim wasn't to simply have one massive texture. The issue was that large textures didn't fit in memory. So it's aim was to improve memory utilisation to be able to load and show only parts of the textures that are visible and hence allow massive textures to be used that otherwise wouldn't fit in memory by streaming in parts (tiles) of that texture when required.
that would load in and out of memory based on what is visible to the player by having this feedbackStream. Read the link I sent you.

In fact read the "why feedback?" Section of the MS sampler feedback link you have sent says exactly that as to why Sampler feedback.

Since you need to aggressively optimize for memory, you bind resources to cope with the demand for different LODs. Perhaps you use a texture streaming system; perhaps it uses tiled resources to keep those gigantic 4K mip 0s non-resident if you don’t need them.

Sound familiar? That's megatextures aim. Now how does it improve on it is the main question.

Texture sampling hardware was extended to allow writing back what was sampled. Previously texture sampling hardware could not do this. This was clearly stated in the video I linked.

Please elaborate what you mean here because we have shader core feedback. What is 'texture sampling hardware'?
At the least please timestamp where this hardware is mentioned.

What you are describing is DirectX just exposing part of something that was available in hardware already. In fact once RDNA2 comes out and they release their latest drivers for it there is talk about whether previous cards will get this update for DX12U like the 2018 RTX cards because these are not new hardware features they are DX API features. (it won't, not because of SF but because DX12U has RT in the feature level hence why cards from 2018 that had raytracing got it already and AMD won't update their old GPUs.)

Look up Feedback Rendering in this paper and tell me what is done on top of that.

https://dl.acm.org/doi/pdf/10.1145/2343483.2343488

The only reason this is "the new hotness" is because of SSDs making this much more useful not a GPU hardware feature that didn't make this possible before.

 

[rnlval]

Yes as stated.
I've read somewhere textures are the biggest chunk thats constantly loaded on hdds and about 25% of all data thats usually loaded.
However this was last gen. I have no idea how things will be on next-gen. Hard to tell.

Either way this example was all in favor of xsx anyways. My guess is PS5's speed advantage will be bigger in real life environments if its neccesary. Because the 60% compression of bcpqck will only effect 25% of the overall data thats beeing loaded.
The rest will use zlip which has about 25% less compression then kraken.

Xsx: 100GB Data = 25GB Textures @ 60% Compression + 75GB Other Data @ 22.5% Compression = 10GB compressed Textures + 58.125 GB other Data = 68.125GB @ 2.4GB/s = 28.39seconds

PS5: 100GB @ 30% compression = 70GB @ 5.5GB/s = 12.73 seconds

Just another take at this. Theres no way we can say which scenario games will use in the future. Either way PS5's SSD will always be faster. Even in the worst case its still 24% faster then xsx.
In this case here its even about 55% faster.

Define this "75GB Other Data".

 

[Trueblakjedi]

1. That .5 difference was ~40%, the current 1.8 is ~15%, if math serves me right

2. More FUD. This has been explained tons of times by people that have way more insight on the tech behind the clock speed.

3. If you're talking about current game design I wholeheartedly agree. But if you start leveraging the SSD for new ways of designing your games, you will absolutely see a difference (in first party games)

4. Nobody skimped on GPU. The faster Xbox knuckleheads understand that both consoles were designed differently, the faster you'll realize that both of them will perform similarly in third party games.

Its a 44% difference in CU.. thats nothing to sneeze at.

 

[Metroiddarks]

As I pointed out, texture streaming is not seldom used. It is common. Just google “[engine name] texture streaming”. The streaming is the same as what you’re talking about - moving mip levels or tiles in and out of memory.



1) Texture sampling hardware was extended to allow writing back what was sampled. Previously texture sampling hardware could not do this. This was clearly stated in the video I linked.

2) Not sure what “not show on new hardware” means. Do you mean why it might not be on the PS5? Because it’s a brand new feature for AMD and it’s not out of the question that it isn’t in their RDNA implementation (And other stated reasons).



Yes, in order to cover terrain with a massive texture:

id Tech 4 - Wikipedia

en.wikipedia.org

”The MegaTexture technology tackled this issue by introducing a means to create expansive outdoor scenes. By painting a single massive texture (32,768×32,768 pixels, though it has been extended to larger dimensions in recent versions of the MegaTexture technology) covering the entire polygon map and highly detailed terrain, the desired effects can be achieved.”

Wow Wikipedia always so complet, you never think how they accomplish to use those huge textures in that year.

ID Software introduce the concept of Virtual Texture Render because they had a problem of the memory available in consoles. So in order to
improve it they create a solution capable of just take an small section of the corresponding mip level and send to physical memory in real time.

http://mrelusive.com/publications/papers/Software-Virtual-Textures.pdf
This solution exists in almost the engines now, Microsoft just improve its last solution which is okay.

Another example of Ubisoft which takes the same technique an improved for its purposes, please don't tell me than you actually think we still load all those texture even if are no complete visible.

http://twvideo01.ubm-us.net/o1/vault/gdc2015/presentations/Chen_Ka_AdaptiveVirtualTexture.pdf

For example here Microsoft talk of the revolutionary technology of 2020 almost seven years ago, weird right

Sample with Feedback sounds familiar right

 

[OptimistPrime]

Please elaborate what you mean here because we have shader core feedback. What is 'texture sampling hardware'?
At the least please timestamp where this hardware is mentioned.

So you want me to read this paper and you won’t even watch this relatively short video that explains everything?

Look up Feedback Rendering in this paper and tell me what is done on top of that.

It’s pretty clear that this is just a mix up of terminology. Just because you see the word “feedback” doesn’t make it the same. Feedback from what? As described in this paper, the feedback has to be rendered separately. Sampler Feedback is just writing back data that’s already there in normal texturing. The video I mentioned contrasts these techniques to what Sampler Feedback gives you cheaply and supposedly more accurately.

...

please don't tell me than you actually think we still load all those texture even if are no complete visible.

...

Sample with Feedback sounds familiar right

1) That slide literally talks about 128k textures. That’s not the same kind of problem as more general texture streaming. We’re talking more like 4K textures. Yes we’re talking about the same hardware features, but the whole point I tried to differentiate is because Three claimed these technologies are seldom used, which I disagreed with.

2) You clearly have not been reading my posts carefully. I’m the one arguing that texture streaming is common!

3) “Sample with Feedback” This is not the same as Sampler Feedback. Listen carefully to how he describes this new feature in 2012, then watch the dx12 video on Sampler Feedback. It’s pretty clear.

 

[itsjustJEFF]

This shit is still going on?!?! You know what this thread really shows me... how much better Xbox is at messaging than PS is. Both these guys knew what the other was doing. Xbox got ahead of PS and talked about virtual memory, to "match" the SSD advantages of PS5, but it's just not the same. But in the end, this little advantage here and there will basically make them equal. So really, all the fighting is just a waste.

 

[MasterCornholio]

But in the end, this little advantage here and there will basically make them equal.

The I/O system isn't equal though. One of them is faster than the other.

 

[Metroiddarks]

1) That slide literally talks about 128k textures. That’s not the same kind of problem as more general texture streaming. We’re talking more like 4K textures. Yes we’re talking about the same hardware features, but the whole point I tried to differentiate is because Three claimed these technologies are seldom used, which I disagreed with.

Of course the size will be different is a game of more than 10 years ago that is not the point.

My claim is SFS is no at unique tecnologie without parallel which can allow to save more than 3 times the memory compare to ther actual solution in the market.

Yes could be better but not so much.

2) You clearly have not been reading my posts carefully. I’m the one arguing that texture streaming is common!

But you claim this as goal of megatexture

"order to cover terrain with a massive texture:"

That is not true completely.

3) “Sample with Feedback” This is not the same as Sampler Feedback. Listen carefully to how he describes this new feature in 2012, then watch the dx12 video on Sampler Feedback. It’s pretty clear.

Sampler Feedback is a evolution of Tiled Resources that was my point.

Before the devs doesn't have much control using things Tiled resource (was ) so they don't in a easy way from which mip was taken that specific texture
that give more control than before of which mips is load and which part, this for example allow to them if they want to save some lighting computations
above that texture, that sounds great but still be an evolution for me.

Is good feature but is not magic because as they said you cannot just called every time they want, that is why they have to appeal to a stochastic discard
they mention two 2 number one discard 99% and the other one out of 128 times for a good performance tradeoff.

 

[Dee_Dee]

The event in June is supposed to be hardware related right? That would be a good time for MS to actually show what XVA can do. Hoping they do that.

 

[ToadMan]

What would be a realistic expectation as far as GPU boost, an even 2.0Ghz?

At this stage it’s not realistic to expect any boost. Or maybe “realistic” is the wrong way to say it - it’s more that there doesn’t seem to be anything to gain by increasing clock speed.

The problem is power. The Xsex gpu is rated to max 200W.

A similar PC Card - 2080rtx - that’s rated to 250W. The Xsex is already running their gpu at a significantly lower power than the equivalent PC Card (this is usual for consoles).

The xb1 gpu had a power rating of 95W. That gpu was roughly equivalent to a GTX750 which had a power rating of 55W.

The XB1 had a power surfeit for its gpu and the cooling was good enough, so MS could afford to increase the clock without risking system stability.

Xsex isn’t enjoying this scenario and that’s why it won’t get a clock boost.

 

[ToadMan]

They added 50 MHz on the the Xbox One. With the cooling they have the XSX adding 150MHz doesn’t sound too bad.

Cooling is only half the problem. There needs to be enough power.

Xb1 had a power surfeit compared to the equivalent PC Card so it could take a clock increase.

Xsex doesn’t have a similar power surfeit. There won’t be a clock increase [except] with a redesign.

 

[ToadMan]

The Xbox One had a BIOS update after launch. No way the PS5 is increasing any clocks. They already need to use AMD shift to hit 2.23GHz on the GPU.

You are confused about the technology.

The limit for both PS5 and Xsex is power. Clock speeds are not the fundamental limit any more than tflops are.

Work done by the processor (% of max tflop) during a single clock tick determines the power draw.

If the clock goes up, the power draw goes up for constant work load.

If the work load goes up, the power draw goes up for constant clock.

Xsex has a GPU discrete power limit. When that limit is reached, there is nowhere for the gpu to go. It either drops frames (reduces work) or keeps going until it hits the thermal limit and shuts down altogether. It may be the cpu is underutilising it’s power at this time - it doesn’t matter, the gpu limit is reached and the system reacts.

PS5 has a shared cpu/gpu power limit. When the gpu reaches its power limit there is the option of allocating unused powered from the cpu to allow the gpu to keep its workload.

Reducing the PS5 clock only occurs if the system total power budget is allocated - that would mean the cpu and gpu are already at their max work load (for the available power). This is an unusual scenario in practice.

 

[ToadMan]

It does work that way. Percentages are just a statistical measurement and you know what they say about statistics...

A 2 TFLOP difference is greater than a .5 TLOP difference. It will show in the games.

If you’re going to use metrics like tflops and clock speeds, percentage comparison is exactly how it works.

PS4 had 50% more CUs clocked 7% slower than x1.

The PS4 achieved the same FPS as x1 at one step higher resolution. The vast majority of users wouldn’t spot the difference in a side by side test.

This generation the Xsex has 40% more CUs running 20% slower than PS5.

There won’t be any difference is res or FPS for all practical intents this gen.

 

[Kagey K]

This shit is still going on?!?! You know what this thread really shows me... how much better Xbox is at messaging than PS is. Both these guys knew what the other was doing. Xbox got ahead of PS and talked about virtual memory, to "match" the SSD advantages of PS5, but it's just not the same. But in the end, this little advantage here and there will basically make them equal. So really, all the fighting is just a waste.

Looks like you are trying to downplay the actual advantages.

There isn’t going to be much “matching” or small differences going on.

I don’t have anything else to say here until the games start proving it.

I think some people are about to get woken up though.

 

[rnlval]

At this stage it’s not realistic to expect any boost. Or maybe “realistic” is the wrong way to say it - it’s more that there doesn’t seem to be anything to gain by increasing clock speed.

The problem is power. The Xsex gpu is rated to max 200W.

A similar PC Card - 2080rtx - that’s rated to 250W. The Xsex is already running their gpu at a significantly lower power than the equivalent PC Card (this is usual for consoles).

The xb1 gpu had a power rating of 95W. That gpu was roughly equivalent to a GTX750 which had a power rating of 55W.

The XB1 had a power surfeit for its gpu and the cooling was good enough, so MS could afford to increase the clock without risking system stability.

Xsex isn’t enjoying this scenario and that’s why it won’t get a clock boost.

From https://www.techpowerup.com/review/nvidia-geforce-rtx-2080-founders-edition/31.html

RTX 2080 FE has peak 226 watts and 215 watts for average gaming.

 

[ToadMan]

1. The difference magnitude remains similar.
GPU's FLOPS
X1X vs PS4 Pro = 1.8 TFLOPS GCN
XSX vs PS5 = 1.872‬ TFLOPS RDNA

PS4 Pro to PS5 TFLOPS increase = 6.075‬ TFLOPS
X1X to XSX TFLOPS increase = 6.147‬ TFLOPS

GPU's memory bandwidthfor render targets
X1X vs PS4 Pro = 102 GB/s
XSX vs PS5 = 112 GB/s

PS4 Pro to PS5 memory bandwidth increase = 224 GB/s
X1X to XSX memory bandwidth increase = 234 GB/s

The percentage argument hides the actual magnitude difference.

The next-generation consoles shifted to a higher level with similar gains when compared X1X and PS4 Pro respectively.

The magnitude difference is irrelevant because the workload (fidelity of graphics, Size of data etc) has increased by a similar magnitude.

Percentage comparisons are the only valid way to compare the systems.

 

[ToadMan]

From https://www.techpowerup.com/review/nvidia-geforce-rtx-2080-founders-edition/31.html

RTX 2080 FE has peak 226 watts and 215 watts for average gaming.

Wel the the Xsex is already close to its power limit. As I mentioned - there won’t be a clock increase.

 

[Kagey K]

Wel the the Xsex is already close to its power limit. As I mentioned - there won’t be a clock increase.

And if there is who cares?

The SSD will make up for it.

Right?

 

[rnlval]

The magnitude difference is irrelevant because the workload (fidelity of graphics, Size of data etc) has increased by a similar magnitude.

Percentage comparisons are the only valid way to compare the systems.

The magnitude difference is relevant since it didn't disappear.

Wel the the Xsex is already close to its power limit. As I mentioned - there won’t be a clock increase.

My argument is not about XSX GPU clock speed increase potential. My argument to correct your RTX 2080's 250 watts number.

 

[ToadMan]

And if there is who cares?

The SSD will make up for it.

Right?

No because the SSD is a feature that impacts game design and storage management leading to a consequent positive effect on fidelity.

The SSD positively impacts the performance of the system as a whole - sound, graphics, game design, O/S.

The gpu is focused purely on graphics.

If you read up on computer system architecture you’ll find its not difficult to comprehend how all these elements interact to achieve performance.

 

[rntongo]

You are confused about the technology.

The limit for both PS5 and Xsex is power. Clock speeds are not the fundamental limit any more than tflops are.

Work done by the processor (% of max tflop) during a single clock tick determines the power draw.

If the clock goes up, the power draw goes up for constant work load.

If the work load goes up, the power draw goes up for constant clock.

Xsex has a GPU discrete power limit. When that limit is reached, there is nowhere for the gpu to go. It either drops frames (reduces work) or keeps going until it hits the thermal limit and shuts down altogether. It may be the cpu is underutilising it’s power at this time - it doesn’t matter, the gpu limit is reached and the system reacts.

PS5 has a shared cpu/gpu power limit. When the gpu reaches its power limit there is the option of allocating unused powered from the cpu to allow the gpu to keep its workload.

Reducing the PS5 clock only occurs if the system total power budget is allocated - that would mean the cpu and gpu are already at their max work load (for the available power). This is an unusual scenario in practice.

You're the one that has these things confused.

The PS5 uses a variable clock rate on its CPU and GPU.

How it works in the Devkits:
The Devkits have one fixed power profile so Devs determine what they want to maximize using AMD shift. According to Rich at DF, Devs have throttled the CPU to pour power into the GPU to ensure a consistent 2.23GHz.


How it works in the Retail consoles:
There is a fixed power profile per workload which creates a range of variable workloads in terms of CPU/GPU performance. The clock rate for each processor is determined by workload. The APU is able to determine through an algorithm which workload is GPU and CPU intensive. In the case that it's GPU intensive, power is shifted from the CPU to the GPU to ensure a consistent 2.23GHz and reverts back to a lower clock when not GPU intensive.

It's all explained here:
PlayStation 5 New Details From Mark Cerny: Boost Mode, Tempest Engine, Back Compat + More

 

[ToadMan]

The magnitude difference is relevant since it didn't disappear.


My argument is not about XSX GPU clock speed increase potential. My argument to correct your RTX 2080's 250 watts number.

Actually the magnitude performance had disappears when one applies it to next generation data loads.

If you want to run last gen games on your new machine then you could have a point. I’m sure the majority of people want to play newly designed games at the highest levels of fidelity.


As for Gou power - Tom’s has actual power recess for both the 2080 and the 5700 - they both pros up to about 250W. Sorry.

 

[rntongo]

Cooling is only half the problem. There needs to be enough power.

Xb1 had a power surfeit compared to the equivalent PC Card so it could take a clock increase.

Xsex doesn’t have a similar power surfeit. There won’t be a clock increase [except] with a redesign.

How do you know that? What would be the additional amount of power needed? It doesn't sound true.

 

[Kagey K]

No because the SSD is a feature that impacts game design and storage management leading to a consequent positive effect on fidelity.

The SSD positively impacts the performance of the system as a whole - sound, graphics, game design, O/S.

The gpu is focused purely on graphics.

If you read up on computer system architecture you’ll find its not difficult to comprehend how all these elements interact to achieve performance.

Oh, I didn’t realize. Are these Sony specific features?

Or does the Series X SSD also affect those same things, before over clocking?

 

[rnlval]

If you’re going to use metrics like tflops and clock speeds, percentage comparison is exactly how it works.

PS4 had 50% more CUs clocked 7% slower than x1.

The PS4 achieved the same FPS as x1 at one step higher resolution. The vast majority of users wouldn’t spot the difference in a side by side test.

This generation the Xsex has 40% more CUs running 20% slower than PS5.

There won’t be any difference is res or FPS for all practical intents this gen.

There would be differences with resolution and FPS just like RTX 2070 at 2230Mhz OC vs RTX 2080 Super.

 

[ToadMan]

You're the one that has these things confused.

The PS5 uses a variable clock rate on its CPU and GPU.

How it works in the Devkits:
The Devkits have one fixed power profile so Devs determine what they want to maximize using AMD shift. According to Rich at DF, Devs have throttled the CPU to pour power into the GPU to ensure a consistent 2.23GHz.


How it works in the Retail consoles:
There is a fixed power profile per workload which creates a range of variable workloads in terms of CPU/GPU performance. The clock rate for each processor is determined by workload. The APU is able to determine through an algorithm which workload is GPU and CPU intensive. In the case that it's GPU intensive, power is shifted from the CPU to the GPU to ensure a consistent 2.23GHz and reverts back to a lower clock when not GPU intensive.

It's all explained here:
PlayStation 5 New Details From Mark Cerny: Boost Mode, Tempest Engine, Back Compat + More

Your text is almost correct. The inaccuracy is that you are still implying Clock speed is the driving parameter. Clock speed is a resultant parameter with this paradigm.

Clock speeds are not driving the algorithm - power consumption is.

If the software runs within the Total Power budget, the cpu and gpu will remain at 100% clock continuously.

When the TOTAL power budget (note total, not gpu or cpu discrete power budgets) is exceeded the system will declock.

Xsex has identical limitations - there is a discrete power budget for its cpu and a discrete power budget for its gpu. What Xsex lacks is the dynamic ability to allocate power between the cpu and gpu. Devs don’t get a choice of how they allocate power they work to the gpu and cpu power constraints - if they hit the power limit for the gpu, they’ll drop frames even if the cpu is idle at that moment and the power is available to drive more gpu work.

 

[rnlval]

Actually the magnitude performance had disappears when one applies it to next generation data loads.

If you want to run last gen games on your new machine then you could have a point. I’m sure the majority of people want to play newly designed games at the highest levels of fidelity.


As for Gou power - Tom’s has actual power recess for both the 2080 and the 5700 - they both pros up to about 250W. Sorry.

Nvidia GeForce RTX 2080 Founders Edition Review: Faster, More Expensive Than GeForce GTX 1080 Ti

Nvidia's GeForce RTX 2080 Founders Edition is in a difficult position. It has to compete against GeForce GTX 1080 Ti at a higher price point, and with a number of features not yet available to enjoy. Can it shine as bright as the flagship RTX 2080 Ti?

www.tomshardware.com

According to toms hardware, RTX 2080 FE has 225.5 watts average for gaming and 224 watts average for the torture test.

Magnitude performance doesn't disappear with next-gen workloads just as RTX 2070 OC at 2230 Mhz wouldn't beat RTX 2080 Super. AMD hasn't released RTX 2080 level SKU for NAVI.

 

[ToadMan]

Oh, I didn’t realize. Are these Sony specific features?

Or does the Series X SSD also affect those same things, before over clocking?

The Xsex won’t be overclocked for the reasons I gave.

Yes the Xsex SSD offers a positive Impact on these features albeit at a reduced level to that of the PS5 based on current information.

Perhaps you should read a little about these machines. Your questions imply you haven’t done much research.

 

[ToadMan]

Nvidia GeForce RTX 2080 Founders Edition Review: Faster, More Expensive Than GeForce GTX 1080 Ti

Nvidia's GeForce RTX 2080 Founders Edition is in a difficult position. It has to compete against GeForce GTX 1080 Ti at a higher price point, and with a number of features not yet available to enjoy. Can it shine as bright as the flagship RTX 2080 Ti?

www.tomshardware.com

According to toms hardware, RTX 2080 FE has 225.5 watts average for gaming and 224 watts average for the torture test.

Magnitude performance doesn't disappear with next-gen workloads just as RTX 2070 OC at 2230 Mhz wouldn't beat RTX 2080 Super. AMD hasn't released RTX 2080 level SKU for NAVI.

I wansn’t using the FE version as my comparison point.

But I’m not sure what you are getting at with this comparison anyway. It simply serves to confirm that the Xsex won’t be receiving any up clock because it is near the max power rating.

Which was the original point I answered.

 

[Panajev2001a]

Looks like you are trying to downplay the actual advantages

Said it in a thread when suddenly* the SSD difference disappeared thanks to a mysterious 2-3x bandwidth multiplier that came out of nowhere... really ?!

*after the argument “oh, yeah you are going to go from 2s to 1s loading times” did not stick and UE5 demo put the fear that there may be more to it

 

[Kagey K]

The Xsex won’t be overclocked for the reasons I gave.

Yes the Xsex SSD offers a positive Impact on these features albeit at a reduced level to that of the PS5 based on current information.

Perhaps you should read a little about these machines. Your questions imply you haven’t done much research.

Id prefer to see them in action, if you want to make factual statements about them I’d hope you could provide me with the insight before the games come out.

The Series X is going to be stronger in every way except the SSD. It has a better GPU, better CPU and better integrated audio. It also has an SSD.

Yet somehow you want people to think the PS5 is better, and when they ask you why, you say “figure it out”

It seems it’s not just Sony who can’t figure out their messaging.

 

[rnlval]

I wansn’t using the FE version as my comparison point.

But I’m not sure what you are getting at with this comparison anyway. It simply serves to confirm that the Xsex won’t be receiving any up clock because it is near the max power rating.

Which was the original point I answered.

The 1st Toms hardware's RTX 2080 review from https://www.tomshardware.com/topics/graphics/reviews/page/3 is RTX 2080 FE.

AIB RTX 2080 factory OC has higher power consumption.

 

[Panajev2001a]

and better integrated audio

Whoa... amongst the usual laundry / bragging list and usual downplaying of PS5... what is this doing here mate ?

 

[rnlval]

Your text is almost correct. The inaccuracy is that you are still implying Clock speed is the driving parameter. Clock speed is a resultant parameter with this paradigm.

Clock speeds are not driving the algorithm - power consumption is.

If the software runs within the Total Power budget, the cpu and gpu will remain at 100% clock continuously.

When the TOTAL power budget (note total, not gpu or cpu discrete power budgets) is exceeded the system will declock.

Xsex has identical limitations - there is a discrete power budget for its cpu and a discrete power budget for its gpu. What Xsex lacks is the dynamic ability to allocate power between the cpu and gpu. Devs don’t get a choice of how they allocate power they work to the gpu and cpu power constraints - if they hit the power limit for the gpu, they’ll drop frames even if the cpu is idle at that moment and the power is available to drive more gpu work.

From



Without AMD's Smartshift technology,

• GPU's 2 Ghz was almost unreachable with a fixed frequency strategy.
• CPU was running at 3 Ghz with a fixed frequency strategy.

Mark Cerny has revealed fixed frequencies for CPU and GPU for PS5. LOL.


XSX's fixed frequency strategy has CPU 16 threads mode at 3.6Ghz or 8 threads mode at 3.8 Ghz and GPU at 1825 Mhz.

XSX has a clock speed boost mode feature when CPU has 8 threads.

MS limits clock speed boost mode for XSX CPU but APU's boost mode features are still on the machine.

 

[Panajev2001a]

From


Without AMD's Smartshift technology,

• GPU's 2 Ghz was almost unreachable with a fixed frequency strategy.
• CPU was running at 3 Ghz with a fixed frequency strategy.

Mark Cerny has revealed fixed frequencies for CPU and GPU for PS5. LOL.

XSX's fixed frequency strategy has CPU 16 threads mode at 3.6Ghz or 8 threads mode at 3.8 Ghz and GPU at 1825 Mhz.

XSX has a clock speed boost mode feature when CPU has 8 threads.

MS limits clock speed boost mode for XSX CPU but APU's boost mode features are still on the machine.

He was talking about early initial strategy not what they had two seconds before and so they decided to flick the SmartShift switch (part of their variable/unlocked clocks strategy not the only bit as part of the video).
Then they decided to focus on fixing power consumption target and optimise around that. You are being very sensationalistic there. What they have now is a system that allows you to trade off workload and workload complexity for frequency in some cases and trade power across units when one needs some help and the other can spare cycles: on top of it, if needed slight down clocking (with voltage reduction too likely) allowing minimal performance impact on the unit with a sizeable power consumption reduction)
All with tooling which makes it easy to optimise for it.

Take those statements in context of console development when it started and the box and cooling they thought about giving the console as well as take CPU and GPU together not split as if they were different chips with their own cooling (also the SSD I/O is a considerable source of heat they need to account for).
The big focus on the GPU was always very high frequency and to design that regardless of the workload and not set a low baseline for the CPU. Thoughts you need to have if you want to prevent a preventable RRoD like scenario.

Nothing that hidden or mysterious, but sure it does not play well with the idea that 10.23 TFLOPS is fake and 9.2 TFLOPS is the real number now, does it ?

 

[ToadMan]

Id prefer to see them in action, if you want to make factual statements about them I’d hope you could provide me with the insight before the games come out.

The Series X is going to be stronger in every way except the SSD. It has a better GPU, better CPU and better integrated audio. It also has an SSD.

Yet somehow you want people to think the PS5 is better, and when they ask you why, you say “figure it out”

It seems it’s not just Sony who can’t figure out their messaging.

I didn’t say “figure it out” - I was pointing out there is no capacity to over clock the Xsex as happened with xb1. I explained that the power is the limiting factor.

You then decided to talk about the SSD and I agreed to converse in simplistic terms with you on that.

Now you want to raise other power figures that have nothing to do with the original point.

That why I suggested you go and do your own research. I’m not here to give you a technology lesson.

 

[ToadMan]

The 1st Toms hardware's RTX 2080 review from https://www.tomshardware.com/topics/graphics/reviews/page/3 is RTX 2080 FE.

AIB RTX 2080 factory OC has higher power consumption.

https://www.tomshardware.com/reviews/nvidia-geforce-rtx-2080-super-turing-ray-tracing,6243-4.html

https://www.tomshardware.com/reviews/amd-radeon-rx_5700-rx_5700_xt,6216-5.html

Hope that helps

 

[ToadMan]

From



Without AMD's Smartshift technology,

• GPU's 2 Ghz was almost unreachable with a fixed frequency strategy.
• CPU was running at 3 Ghz with a fixed frequency strategy.

Mark Cerny has revealed fixed frequencies for CPU and GPU for PS5. LOL.


XSX's fixed frequency strategy has CPU 16 threads mode at 3.6Ghz or 8 threads mode at 3.8 Ghz and GPU at 1825 Mhz.

XSX has a clock speed boost mode feature when CPU has 8 threads.

MS limits clock speed boost mode for XSX CPU but APU's boost mode features are still on the machine.

I think you replied to wrong post ...

 

[Three]

So you want me to read this paper and you won’t even watch this relatively short video that explains everything?



It’s pretty clear that this is just a mix up of terminology. Just because you see the word “feedback” doesn’t make it the same. Feedback from what? As described in this paper, the feedback has to be rendered separately. Sampler Feedback is just writing back data that’s already there in normal texturing. The video I mentioned contrasts these techniques to what Sampler Feedback gives you cheaply and supposedly more accurately.




1) That slide literally talks about 128k textures. That’s not the same kind of problem as more general texture streaming. We’re talking more like 4K textures. Yes we’re talking about the same hardware features, but the whole point I tried to differentiate is because Three claimed these technologies are seldom used, which I disagreed with.

2) You clearly have not been reading my posts carefully. I’m the one arguing that texture streaming is common!

3) “Sample with Feedback” This is not the same as Sampler Feedback. Listen carefully to how he describes this new feature in 2012, then watch the dx12 video on Sampler Feedback. It’s pretty clear.

I watched the entire video. I want to know what you're referring to. I mentioned which section of the paper to read.

How is the feedback different. It has to be rendered separately? Look at the DX12 spec on SF. In fact watch your own video at 11:13. Then start reading the section in the paper I linked.

This isn't a terminology mixup. I want to know what hardware you think is now there for sampler feedback. You said something that is not hardware related and I gave you an example of it already being done.

Edit: I decided make life easier for you and just post the section of the paper here:

While a sparse representation makes it possible to render with a partially resident texture,
feedback is necessary for determining which parts of the texture need to be resident. Texture
feedback needs to be rendered to a separate buffer to store the virtual page coordinates
(x,y), desired mip level, and virtual texture ID (to allow multiple virtual textures). This
information is then used to pull in the texture pages needed to render the scene. The
feedback can be rendered in a separate rendering pass or to an additional render target during
an existing rendering pass. An advantage of rendering the feedback is that the feedback is
properly depth tested, so the virtual texture pipeline is not stressed with requests for texture
pages that are ultimately invisible. When a separate rendering pass is used it is fine for the
feedback to be rendered at a significantly lower resolution (say 10x smaller).

 

[Ar¢tos]

Said it in a thread when suddenly* the SSD difference disappeared thanks to a mysterious 2-3x bandwidth multiplier that came out of nowhere... really ?!

*after the argument “oh, yeah you are going to go from 2s to 1s loading times” did not stick and UE5 demo put the fear that there may be more to it

The difference didn't disappear, it reversed.
XSX ssd started at 4.8 gb/s and now it's at 7gb/s. It gets faster every day. By the time it releases it will be at 50gb/s.

Meanwhile, ps5 SSD started at 9gb/s and now it's at 4.5gb/s.
Because apparently the SSD it's bottlenecked by the GPU that caps at 4.5gb/s.
Stupid Sony spent tons in R&D creating the SSD and its custom controller and never realized that their weak GPU couldn't handle more than 4.5gb/s from the SSD.

 

[Dodkrake]

1. The difference magnitude remains similar.
GPU's FLOPS
X1X vs PS4 Pro = 1.8 TFLOPS GCN
XSX vs PS5 = 1.872‬ TFLOPS RDNA

PS4 Pro to PS5 TFLOPS increase = 6.075‬ TFLOPS
X1X to XSX TFLOPS increase = 6.147‬ TFLOPS

GPU's memory bandwidthfor render targets
X1X vs PS4 Pro = 102 GB/s
XSX vs PS5 = 112 GB/s

PS4 Pro to PS5 memory bandwidth increase = 224 GB/s
X1X to XSX memory bandwidth increase = 234 GB/s

The percentage argument hides the actual magnitude difference.

The next-generation consoles shifted to a higher level with similar gains when compared X1X and PS4 Pro respectively.

Its a 44% difference in CU.. thats nothing to sneeze at.

God, it's always the same people in a circular argument. Not even gonna try anymore, you want other posters to just lose their shit and get banned, I won't. Good day

 

[rntongo]

Your text is almost correct. The inaccuracy is that you are still implying Clock speed is the driving parameter. Clock speed is a resultant parameter with this paradigm.

Clock speeds are not driving the algorithm - power consumption is.

If the software runs within the Total Power budget, the cpu and gpu will remain at 100% clock continuously.

When the TOTAL power budget (note total, not gpu or cpu discrete power budgets) is exceeded the system will declock.

Xsex has identical limitations - there is a discrete power budget for its cpu and a discrete power budget for its gpu. What Xsex lacks is the dynamic ability to allocate power between the cpu and gpu. Devs don’t get a choice of how they allocate power they work to the gpu and cpu power constraints - if they hit the power limit for the gpu, they’ll drop frames even if the cpu is idle at that moment and the power is available to drive more gpu work.

The power budget is per workload and in order for the GPU to hit the max clocks they have to draw power The CPU. We have concrete proof of this in the dev kits and its what some devs have been doing.

Why are devs throttling the CPU to consistently get 2.23GHz on the GPU?

 

[Panajev2001a]

God, it's always the same people in a circular argument. Not even gonna try anymore, you want other posters to just lose their shit and get banned, I won't. Good day

It is about trying to make the narrative the norm or change the topic and/or just get the other fans pissed off for some of the hardcore fans.

Of course people will cycle through talking points without caring much about being right or wrong, once they pushed it far enough they will just move to the next one and then come back to it at some other point as if nothing happened/they got no pushback. Hopefully people landing on the posts seeing this said over and over will go from “what the heck are they saying?” To a most believing there must be something behind it after all... beauty of repetition :/...

 

[Panajev2001a]

The power budget is per workload and in order for the GPU to hit the max clocks they have to draw power The CPU. We have concrete proof of this in the dev kits and its what some devs have been doing.

Do you have links to all the concrete and I suspect voluminous proofs?

 

[rntongo]

Do you have links to all the concrete and I suspect voluminous proofs?

between 4:40-5:00 minutes, Rich explains devs have been throttling the CPU to get more power to the GPU in order to hit 2.23GHz. Later on I think 5:40-6:00 he explains in the retail units their will be multiple workloads to automatically detect gpu and cpu intensive tasks and adjust the clocks depending on the workload. Inorder to hit 2.23GHz the CPU has to transfer power to the GPU and get a reduction in its clockspeed.
DF Direct: PlayStation 5 - The Official Specs, The Tech + Mark Cerny's Next-Gen Vision

 

[Panajev2001a]

between 4:40-5:00 minutes, Rich explains devs have been throttling the CPU to get more power to the GPU in order to hit 2.23GHz. Later on I think 5:40-6:00 he explains in the retail units their will be multiple workloads to automatically detect gpu and cpu intensive tasks and adjust the clocks depending on the workload. Inorder to hit 2.23GHz the CPU has to transfer power to the GPU and get a reduction in its clockspeed.
DF Direct: PlayStation 5 - The Official Specs, The Tech + Mark Cerny's Next-Gen Vision

That is not concrete proof from devkit or devs themselves. This is your interpretation of Leadbetter’s interpretation of what he said he read and heard on top of that.

 

[rntongo]

That is not concrete proof from devkit or devs themselves. This is your interpretation of Leadbetter’s interpretation of what he said he read and heard on top of that.

So Rich who interviewed Cerny is no longer credible? Ok What was your interpretation?
Next time I'll find a way to get the devkits to answer your questions.

 

[Panajev2001a]

So Rich who interviewed Cerny is no longer credible? Ok What was your interpretation?
Next time I'll find a way to get the devkits to answer your questions.

You are the one that claimed concrete proof from devkits . Leadbetter is hardly a credible source when he interprets Sony unless he is reporting actual quotes, but his DF history (him and Battaglia) speaks for itself.