Sampler Feedback Streaming will allow Xbox Series X to better the I/O throughput of Sony PS5

Marees

[H]ard|Gawd
Joined
Sep 28, 2018
Messages
2,039
Early this week, Microsoft explained that Sampler Feedback Streaming (SFS) is the key component that will allow Xbox Series X to offer better effective I/O throughput than Sony PS5.

PS5 SSD is faster than Xbox Series X SSD. There is no question about it. But the efficiency provided by Sampler Feedback Streaming tech will allow game developers to enjoy better I/O throughput on Xbox Series X. Consider a game that wants to load 20GB of data.

  • Sony PS5 SSD will take 3.63 seconds to load the data.
  • Theoretically, Xbox Series X SSD will take 8.33 seconds to load the same amount of data. But thanks to SFS, entire 20GB data need not be loaded, only 8GB is required. So, Xbox Series X SSD will load the 8GB data in 3.3 seconds.
  • Even though Xbox Series X loaded 2.5 times less data when compared to Sony PS5, gamers will have the same visual experience on their screens.
https://mspoweruser.com/microsoft-xbox-series-x-i-o-throughput-ps5/

Sampler Feedback Streaming is a brand new method of rendering game textures. They render at far less quality the further away the player is. The resolution then increases the closer the player gets.

Jason Ronald, Director of Program Management on Xbox Series X, wrote an article on Xbox Wire.
Jason details SFS by saying:
“We were able to create and add new capabilities to the Xbox Series X GPU which enables it to only load the sub-portions of a mip level into memory, on-demand, just in time for when the GPU requires the data. This innovation results in approximately 2.5x the effective I/O throughput and memory usage above and beyond the raw hardware capabilities on average”
https://loadingxp.com/xbox-velocity-architecture-explained-by-microsoft/

Speaking in an interview with Spanish site Xataka, Jason Ronald explains:
"Sampler Feedback Streaming (SMS) allows us to load textures and makes the SSD drive act as a multiplier of physical memory that adds to the memory that the machine itself has,”
https://www.techradar.com/in/news/t...l-than-ps5-if-microsoft-copied-sonys-approach

EDIT:

In an interview with Windows Central, Studio Technical Director at The Coalition Mike Rayner commented on the console’s Sampler Feedback Streaming, a new feature in the console’s architect. He says that it will greatly increase the visual detail of worlds:
"We will be exploring how we can use it in future titles to both increase the texture detail in our game beyond what we can fit into memory, as well as reduce load times further by increasing on-demand loading to just before we need it, instead of pre-loading everything up-front as we would use a more traditional ‘level loading’ approach.”
https://gamingbolt.com/xbox-series-...treaming-is-game-changer-says-gears-developer


EDIT 2:

From Eurogamer:

A technique called Sampler Feedback Streaming - SFS - was built to more closely marry the memory demands of the GPU, intelligently loading in the texture mip data that's actually required with the guarantee of a lower quality mip available if the higher quality version isn't readily available, stopping GPU stalls and frame-time spikes. Bespoke hardware within the GPU is available to smooth the transition between mips, on the off-chance that the higher quality texture arrives a frame or two later

https://www.eurogamer.net/articles/digitalfoundry-2020-inside-xbox-series-x-full-specs
 
Last edited:
From AnandTech's deep dive on SSD usage in consoles:

Sampler Feedback Streaming is a GPU feature that makes partially resident textures more useful by allowing shader programs to keep a record of which portions of a texture are actually being used. This information can be used to decide what data to evict from RAM and what to load next—such as a higher-resolution version of the texture regions that are actually visible at the moment.

partially resident textures has been possible on other platforms, but becomes more powerful now. What was originally developed for multi-acre ground textures can now be effectively used on much smaller objects. Sampler feedback allows the GPU to provide the application with more detailed and accurate information about which portions of a texture are actually being displayed. The game can use that information to only issue SSD read requests for those portions of the file. This can be done with a granularity of 128kB blocks of the (uncompressed) texture, which is small enough to allow for meaningful RAM savings by not loading texels that won't be used, while at the same time still issuing SSD reads that are large enough to be a good fit for SSD performance characteristics.

Microsoft has stated that these capabilities add up to the effect of a 2x or 3x multiplier of RAM capacity and SSD bandwidth. I'm not convinced. Sure, a lot of SSD bandwidth can be saved over short timescales by incrementally loading a scene. But I doubt these features will allow the Series X with its ~10GB of VRAM to handle the kind of detailed scenery you could draw on a PC GPU with 24GB of VRAM. They're welcome to prove me wrong, though.

https://www.anandtech.com/show/15848/storage-matters-xbox-ps5-new-era-of-gaming/3
 
SFS is built on top of the Direct X 12 Ultimate/12_2 software feature Sample Feedback
SFS is implemented using a texture filter in the GPU inside Xbox Series X

https://www.anandtech.com/show/15637/microsoft-intros-directx-12-ultimate-next-gen-feature-set

Previously demoed by NVIDIA as texture-space shading, sampler feedback is a broader feature with a few different uses. At a very high level, the idea behind sampler feedback is to allow game engines to track how the texture samplers are being (or will be) used – thus, the samplers give feedback to the engine – allowing the engine to make more intelligent decisions about how the samplers are used and what resources are kept in VRAM.

The principle use case for this, Microsoft envisions, will be in improving texture streaming. By using sampler feedback, game engines can determine what texture tiles are actually going to be needed, and thus only loading up the necessary tiles. This keeps overall VRAM pressure down, ultimately allowing developers to use higher quality textures overall by losing less VRAM to unneeded tiles. Fittingly for the Xbox Series X, this is especially handy when your games are stored on a high speed SSD, as it means the necessary tiles can be pulled in from storage incredibly quickly (almost in a just-in-time fashion), instead of having to stage them in RAM or take measures to mitigate the long access time of a HDD.

TSS_575px.png

Meanwhile texture-space shading is the other major use for this feature. Another efficiency technique, texture-space shading allows for the shading of an object to take place without actually rasterizing it. Microsoft’s example here involves lighting – where an object has its lighting calculated once instead of repeatedly as a rasterized object would require. Ultimately, the central idea behind this feature is to be able to cache and reuse shading results, freeing up GPU resources for other, more important tasks.
 
SFS is implemented using a texture filter in the GPU inside Xbox Series X

This is what James Stanard (@JamesStanard) had to say on Twitter:

Sampler Feedback is one part of SFS. To make it more useful for texture streaming, we added special texture filters that handle when a texture page is not in memory yet. That's custom for XSX.

I believe only "Sampler Feedback" is a guaranteed feature of DX12 Ultimate.

That being said, games have been streaming virtual memory pages for a while. It's called Partially Resident Textures. Using Sampler Feedback to trigger page reads is sort of the "new hotness".

We have custom texture filters that I already mentioned. These are in hardware.

I only said Sampler Feedback was part of DX12 Ultimate. You can use it to figure out what texture pages to stream, but without our custom texture filters, you might notice "pop in" at page boundaries.


https://twitter.com/JamesStanard/status/1250544003524644864?s=20
 
SFS is built on top of the Direct X 12 Ultimate/12_2 software feature Sample Feedback
SFS is implemented using a texture filter in the GPU inside Xbox Series X

https://www.anandtech.com/show/15637/microsoft-intros-directx-12-ultimate-next-gen-feature-set

Previously demoed by NVIDIA as texture-space shading, sampler feedback is a broader feature with a few different uses. At a very high level, the idea behind sampler feedback is to allow game engines to track how the texture samplers are being (or will be) used – thus, the samplers give feedback to the engine – allowing the engine to make more intelligent decisions about how the samplers are used and what resources are kept in VRAM.

The principle use case for this, Microsoft envisions, will be in improving texture streaming. By using sampler feedback, game engines can determine what texture tiles are actually going to be needed, and thus only loading up the necessary tiles. This keeps overall VRAM pressure down, ultimately allowing developers to use higher quality textures overall by losing less VRAM to unneeded tiles. Fittingly for the Xbox Series X, this is especially handy when your games are stored on a high speed SSD, as it means the necessary tiles can be pulled in from storage incredibly quickly (almost in a just-in-time fashion), instead of having to stage them in RAM or take measures to mitigate the long access time of a HDD.


Meanwhile texture-space shading is the other major use for this feature. Another efficiency technique, texture-space shading allows for the shading of an object to take place without actually rasterizing it. Microsoft’s example here involves lighting – where an object has its lighting calculated once instead of repeatedly as a rasterized object would require. Ultimately, the central idea behind this feature is to be able to cache and reuse shading results, freeing up GPU resources for other, more important tasks.

From Microsoft's blog on Sampler Feedback

Sampler Feedback is designed to work well across different GPU hardware implementations. Even if feedback maps are implemented in different ways across various hardware, Direct3D’s exposure of them avoids platform-variation-related burdens on the application developer. Applications can deal with a convenient unified representation of sampler feedback.
Getting Started
To use Sampler Feedback in your application, install the latest Windows 10 Insider Preview build and SDK Preview Build for Windows 10 (20H1) from the Windows Insider Program. You’ll also need to download and use the latest DirectX Shader Compiler. Finally, because this feature relies on GPU hardware support, you’ll need to contact GPU vendors to find out specifics regarding supported hardware and drivers.

You can find more information in the Sampler Feedback specification, located here.
Claire Andrews

https://devblogs.microsoft.com/dire...edback-some-useful-once-hidden-data-unlocked/
 
The Sampler Feedback Streaming patent actually mentions how SFS was designed to accelerate software virtual texturing (like Unreal Engine 5 is using)

From the patent:

"Software-only residency map solutions typically perform two fetches of two different buffers in the shader, namely the residency map and the actual texture map. The primary PRT texture sample is dependent on the results of a residency map sample. These solutions are effective, but require considerable implementation changes to shader and application code, especially to perform filtering the residency map in order to mask unsightly transitions between levels of detail, and may have undesirable performance characteristics. The improvements herein can streamline the concept of a residency map and move the residency map into a hardware implementation."

https://www.reddit.com/r/XboxSeriesX/comments/grbxlr/the_sampler_feedback_streaming_patent_actually/
 
This sounds awesome. Aside from the load speed, this should help a bunch with limited VRAM.

A Reddit post which gives a very good explainer of SFS:

you should be able to use close to the full speed of the SSD for texture streaming.

If the speed of texture loading, after decompression, is around 3.8 GiB/s, that would allow loading 64 MiB per frame at 60fps. This is fast enough to load 1000 tiles each frame, about the equivalent of a single 4096x4096 texture.

For objects approaching from afar, textures will be loaded in when the GPU changes which mipmap it is rendering from, with only a frame or two's delay, so loading should be effectively invisible. If you look at an uncached object up close for the first time, it might render using lower resolution textures for a frame or two. How low that is depends on how much memory the developer wanted to save.

I believe this API requires specific GPU hardware to support, but it is not exclusive to consoles


https://www.reddit.com/r/hardware/comments/fx0bbj/sampler_feedback_streaming_how_games_will_use/
 
This just sounds like a hackish way for the console to partially cover up the deficiencies of limited VRAM.

This could be a good way to squeeze extra performance out of a limited hardware budget but that sort of thing usually introduces other issues(tradeoffs) and is never the cure-all that they market them as.
 
This just sounds like a hackish way for the console to partially cover up the deficiencies of limited VRAM.

This could be a good way to squeeze extra performance out of a limited hardware budget but that sort of thing usually introduces other issues(tradeoffs) and is never the cure-all that they market them as.


Yup, it's just another bandwdith tweak. You're still limited by the same VRAM capacity (only double that of the previous generation.

Meanwhile, high-end PCs are starting to ship with 32GB ram, and they will ,also see a doubling of dedicated VRAM (for a total of 48GB ram)

You want to know why no PC games currently require an SSD? It's because we're inundated with gobs of cache memory. The only thing that is undeniably faster on an SSD is game load times!

I could see SATA SSDs becoming required for newer PC ports, but when you're swimming in memory, you don't need the same m.2 sustained bandwidth...it's going to be at-least another console generation before m.2 becomes require in your gaming PC.
 
Last edited:
Yay. A whole new generation of "pop in" effect.

As the director of Xbox program management says below* the pop-in effect occurs only in GPUs that do not have the hardware based texture filter (which is all GPUs in the market today)

At some point AMD will introduce GPUs with hardware texture filter followed by Nvidia

Sampler Feedback & Partially Resident Textures are already part of DirectX & modern GPUs/games

Sampler Feedback Services is another layer that is sure to come to DirectX on PC once the hardware is ready for that

When SFS is implemented then instead of a popout you would briefly (for 1 or 2 frames), see a low quality texture, until the Full resolution texture is loaded from virtual memory(or disk) to main memory

(This is of course an alternative to the Nanite tech of EPIC games Unreal Engine 5)

*
games have been streaming virtual memory pages for a while. It's called Partially Resident Textures. Using Sampler Feedback to trigger page reads is sort of the "new hotness".

We have custom texture filters that I already mentioned. These are in hardware.

I only said Sampler Feedback was part of DX12 Ultimate. You can use it to figure out what texture pages to stream, but without our custom texture filters, you might notice "pop in" at page boundaries.
 
Sounds cool, but so did the Rapid Packed Math feature of the PS4 Pro's GPU that Sony touted on the PS4 Pro to bridge the gap between it and the Xbone X. I have yet to see that tech advertised or featured in a game either since its announcement a few years ago. I know these are different components and technologies, but I wouldn't put much stock into it until we see the real-world results from it and how it affects cross-platform games on both consoles. As it seems this is a DX12 feature as well and thus exclusive to the Series X, who's to say Sony couldn't implement similar tech into their proprietary libraries (based on OpenGL) to negate this as well?

Personally, specs and tech mean little to me and my decision to get a console, and I'll probably end up with both of these at some point anyways. The PS5 has better games/exclusives and that's all that matters, which is why I have a Switch as well. And any other Xbone games I'll have for PC and can play them in my living room too via my Shield TV/Gamestream or Steam Link.
 
Last edited:
I have yet to see that tech advertised or featured in a game

I expect SFS to be present in Senua's Saga — Hell Blade 2

EDIT:
or maybe not.
It looks like Microsoft is porting Hell Blade 2 to Unreal Engine 5, in which case there may not be a need for implementing SFS in the game !?

The power of Xbox Series X was first demonstrated with the unveiling of Senua’s Saga: Hellblade II. The team will be building the game on Unreal 5 and leveraging the power of Xbox Series X to bring the Hellblade franchise to levels never before seen. The footage shown was captured in-engine and reflects the power of Xbox Series X available to developers to deliver new universes, experiences and games in ways you have never imagined.

https://news.xbox.com/en-us/2020/06/10/everything-you-need-to-know-about-the-future-of-xbox/
 
Last edited:
As the director of Xbox program management says below* the pop-in effect occurs only in GPUs that do not have the hardware based texture filter (which is all GPUs in the market today)

and yet:
For objects approaching from afar, textures will be loaded in when the GPU changes which mipmap it is rendering from, with only a frame or two's delay, so loading should be effectively invisible. If you look at an uncached object up close for the first time, it might render using lower resolution textures for a frame or two. How low that is depends on how much memory the developer wanted to save.

People nitpick about single digit milliseconds. Two frames in most games will be around 33 milliseconds. So as I am panning the view about, things can take 33ms to load from low to high res? That sounds terrible to me and is the very definition of "pop in effect" in games.
 
I think the most important part of this system is that it's coming to Windows too. Combined with DirectStorage, we should be able to effectively match the PS5's storage bandwidth with a good PCI-E 3.0 NVMe drive. So much for the PS5 being a "generation ahead."
People nitpick about single digit milliseconds. Two frames in most games will be around 33 milliseconds. So as I am panning the view about, things can take 33ms to load from low to high res? That sounds terrible to me and is the very definition of "pop in effect" in games.
It's 2 frames to go up one mip size, which is different from current systems that will transition from basically the smallest mip to the largest one rather suddenly, and often far closer to the camera. It should be effectively invisible.
 
If this is only for texture streaming how is it classified as faster than PS5 in I/O? Aren’t there tons of other applicable usage scenarios not related to texture loads?

This to me seems like Sony’s 4K ready machine (PS4 Pro) having to rely on checkerboarding to keep pace visually with Xbox One X. It’s a software implementation to alleviate slow hardware that will have drawbacks.
 
It’s a software implementation to alleviate slow hardware that will have drawbacks.

It is a combination of software & hardware hacks in the series X

The texture filter is a hardware feature built into the GPU in Xbox series X
 
I have yet to see that tech advertised or featured in a game either since its announcement a few years ago. I know these are different components and technologies, but I wouldn't put much stock into it until we see the real-world results from it

This is EuroGamer's comment on Hellblade 2 demo last year:

...Where we veer into 'too good to be true' territory is in level of detail in that there are no visible limits to draw distance and zero evidence of any LOD 'pop-in', something that's very much present and correct in any current-gen game - including the original Hellblade maxed out on PC.
...
The overall asset quality in this trailer is obviously a generation beyond today's games and that includes Hellblade 1 before it. However, the level of its perfection in terms of distant detail - even given the constraints lifted by a 24fps frame-rate and a sub 4K resolution - do seem unrealistic, and I've yet to see research into techniques that could deliver the perfect LOD transitions exhibited in this trailer.
...
If nothing else, we now know that next-gen assets are going to be authored to an incredible level of detail - but I'd say that the level of dynamism and perfection of that detail across distances needs to be demonstrated, while we need to see more content to get a genuine feeling for what new rendering techniques are likely to be in place for the next generation of console hardware...

https://www.eurogamer.net/articles/digitalfoundry-2019-senuas-saga-hellblade-2-trailer-analysis
 
Combined with DirectStorage, we should be able to effectively match the PS5's storage bandwidth with a good PCI-E 3.0 NVMe drive.

You may want to invest in a PC with 32gb Ram & GPU with 10+gb Vram too ...
 

They're not using this Sampler Feedback Streaming tech for that though, or at least it's not mentioned in that article whatsoever, so how is this relevant to Series X and its use of it to negate the PS5's SSD advantage? So like I said, I wouldn't put much stock into this feature until we see it used in real-world cross-platform games to compare it to the PS5 versions when they're here.

Also, it doesn't seem that this tech will help much in terms of initially loading a game up or fast-traveling within a game, and is only good for texture streaming while moving/panning about within a large open-world game, which even with the PS5's marginal SSD advantage, I don't see translating to much faster real-world loading performance over the Series X either, as most cross-platform devs will probably not go out of their way to take advantage of it like I'm sure Sony's first party studios will.
 
This is good stuff, but as was just mentioned, it might not offer a dramatic difference unless a game developer is focused on only one console.

Me? I'm just happy that consoles are about to take some strides forward, and might just drag PCs along in the process (by pushing them toward NVMe SSDs as standard).
 
What are the hardware hacks?

See below text from patent for SFS below:

It mentions that the Residency Map is implemented in hardware (GPU?)

In public, Microsoft have claimed that the texture filter (same as Residency Map !?) is implemented in the GPU.

Without this hardware hack pop-outs will be inevitable when a low resolution texture has to be loaded for the first time.

Otoh, with this hardware hack implemented, the GPU/game is able to display the low resolution image for 1 or 2 frames till the higher resolution texture gets loaded into main memory from virtual memory/disk

"Software-only residency map solutions typically perform two fetches of two different buffers in the shader, namely the residency map and the actual texture map. The primary PRT texture sample is dependent on the results of a residency map sample. These solutions are effective, but require considerable implementation changes to shader and application code, especially to perform filtering the residency map in order to mask unsightly transitions between levels of detail, and may have undesirable performance characteristics. The improvements herein can streamline the concept of a residency map and move the residency map into a hardware implementation."

The Sampler Feedback Streaming patent actually mentions how SFS was designed to accelerate software virtual texturing (like Unreal Engine 5 is using)

From the patent:



https://www.reddit.com/r/XboxSeriesX/comments/grbxlr/the_sampler_feedback_streaming_patent_actually/
 
Not sure why you are calling it a "hack". It is a legitimate new feature supported in hardware as well as DirectX.
 
  • Like
Reactions: T4rd
like this
Interesting speculation by a games developer if Sony has the equivalent of SFS in PS5

He speculates that PS5 has hardware implementation for SF equivalent but SFS equivalent (texture filter) seems to be in software implementation ...


First of all, the XSX is RDNA 2 based, and will be using SFS in hardware,
not software. As you can imagine, doing it natively with custom hardware is extremely well optimised and much faster than software. 4/11
As for PS5? Can it do it at all? Yes, of course. I never mentioned this, as it was obvious to me but clearly not to you. I’m told it’s a mish-mash Od RDNA 1 and 2 with custom bits. HOWEVER, while consoles as old as PS4 supported basic SF in hardware (it’s quite old)... 5/11
...they didn’t support the more current updates to how things are done. HOWEVER (again), this is where things get fuzzy. I’m trying to get a straight answer on if PS5 supports this via hardware or software (shader). 6/11
It’s possible to support this on software based shader, of course, however that would be much much more slower and clearly not be ideal when compared to a dedicated hardware solution. On speaking with... 7/11
... a Principal Graphics Engineer from Sony, he did his best to answer me but is of course bound by NDA. I won’t put a spin on whether his answers made me lean one way or the other, you can read them in the thread and decide for yourself. But he... 8/11
...will try and get back to me today, if NDA allows him to discuss this in public.

So *as of right now*, nothing has changed, we are still in a position of dedicated hardware for XSX, and and software for PS5. In which case, this all still stands. 9/11
I’m going to assume at the very least they have dedicated hardware for the old style SF, which is far less impressive but still good in its own right of course, as this was in current gen consoles. However from what I can gather, Sony may have... 10/11
...taken a different route to Microsoft here.

https://threadreaderapp.com/thread/1284409882083557376.html

via GamingBolt

https://gamingbolt.com/xbox-series-...ng-is-an-absolute-game-changer-says-developer

𝙂𝘼𝙑𝙄𝙉 𝙎𝙏𝙀𝙑𝙀𝙉𝙎 (@Gavavva) Tweeted:
SFS will essentially only load the required mips and totally ignore everything else that would usually need to be called into memory. Why is that important?

Simple.

Because now, that half a gigs worth of uncompressed data, is just 576......kb. 14/14
https://twitter.com/Gavavva/status/1283863698134753281?s=20
 
Hopefully Hunt:Showdown rolls out some DX12 goodness and takes advantage of this "feature" .. pop ups are horrible in that game and one of my few gripes with the game. It's running on DX11 right now.
 
and yet:


People nitpick about single digit milliseconds. Two frames in most games will be around 33 milliseconds. So as I am panning the view about, things can take 33ms to load from low to high res? That sounds terrible to me and is the very definition of "pop in effect" in games.
Yes, but that's a from low res to high res texture, not geometry and if it's going 256x256 to 512x512 it means it's still a little bit aways from you, so the "pop-in" is just going from ever so slightly blurry to more crisp/clear, the transition closest to your view (depending on texture settings of course) is 2k -> 4k up close to your face stuff, but lets be honest, most 2k textures still look pretty great in game, while 4k is for really zoomed in stuff. It's not like geometry LOD where something that wasn't there suddenly appears. I'd like to see it in action before I call it one way or the other, but I don't think this is going to be near as annoying as geometry LOD.
 
SF is on Turing and RDNA2, it is the smarter cousin of prt in the sense that you add a buffer that tracks the texture resident in memory, was it used or not, what level of mip was it, was it the level you wanted or not?


SFS, the unique XSX hardware is a special filter that simplifies the checking of "was this the mip you wanted? Wasn't it?" and automatically queues the command to fetch it from the ssd. This saves you processing time and latency.


The big savings//multiplier effect of bandwidth is already on the PRT, this is still important and unique due to it taking advantage of the top to bottom integration inherent to the console, kudos to the Microsoft engineers! This won't come to pc AFAIK because similar to the cache scrubbers, it requires full hardware integration to have the gpu queue a request to the i/o ignoring the cpu.

WB engineers were talking about this and the ps5 this weekend lamenting the state of the i/o stack native to pc, still they chimed that this still won't matter much until game engines are rebuilt with a way to store memory ready data structures since current consoles make the loading easy, but it is the initial setup of your world, the physics, ai, logic, audio now that which slows you down. (yeap I was bored and stalked tech convos this weekend)
 
Not sure why you are calling it a "hack". It is a legitimate new feature supported in hardware as well as DirectX.
It's a software hack supported by hardware, but so is pretty much all game image rendering, so meh?
 
I could see SATA SSDs becoming required for newer PC ports, but when you're swimming in memory, you don't need the same m.2 sustained bandwidth...it's going to be at-least another console generation before m.2 becomes require in your gaming PC.

I'm not too concerned with the tribal banter of Team Xbox versus Team Playstation versus Team PC. but I what I found very interesting in regards to the Playstation SSD announcement is how games were so fundamentally impacted by the speed of the storage medium. The maps and game play themselves were dictated by the limitations of needing to read data off a slow medium (slow compared to the requirements of loading into memory).

I'm most excited to see how games actually change when the storage bandwidth limitation isn't an issue, or is considerably less of an issue that previous game generations. They were talking about examples of like previous generations having to keep 30 seconds of possible textures in memory, versus now they only need 1 or 2 seconds of possible textures in memory. And the texture can be anywhere on the storage, not placed in strategic locations to address access time issues.

Very cool stuff from a limitation most people probably never thought about.
 
A 30 per cent year-on-year reduction in the cost of memory has dwindled to a mere five per cent, meaning that not only is the silicon more expensive, improvements to memory capacity in line with the 8x boost from Xbox 360 to Xbox One are no longer possible. There's not much Microsoft can do to combat the cost per transistor conundrum - though dedicated hardware for elements like variable rate shading and ray tracing certainly helps - but necessity is the mother of invention, and that's why Microsoft developed shader feedback sampling.

https://www.eurogamer.net/articles/digitalfoundry-2020-xbox-series-x-silicon-hot-chips-analysis

While it's potentially a way to cut prices over time, it may well be the case that it'll open the door to higher capacity console SKUs that don't cost the earth
 
I'm most excited to see how games actually change when the storage bandwidth limitation isn't an issue, or is considerably less of an issue that previous game generations. They were talking about examples of like previous generations having to keep 30 seconds of possible textures in memory, versus now they only need 1 or 2 seconds of possible textures in memory. And the texture can be anywhere on the storage, not placed in strategic locations to address access time issues.

Very cool stuff from a limitation most people probably never thought about.

It may end up being a lot less of an issue that developers are making it out to be. The reason is that while everything they say is true... there's still the limitation of how much texture/object data you can reasonably and affordably develop for a game. So while the technology would allow in theory for a whole lot more variety, in practice that may be too expensive to develop and re-use continues to be the name of the game. Art assets cost time and money, there's only so much you can do reasonably.

What I like to point out for things like that is movies, specifically Wall-e. Both Wall-e and Eve in that movie are fully bespoke designs. Artists designed all the models and textures for each, and gave them unique animations. The rest of the robots, they did not. The developed fundamental building blocks and put those together in different ways to make a bunch of robots that had a similar stylistic theme. Now why do that? It wasn't memory or storage, the movie was rendered on a massive offline cluster. You can have as complex a scene as you like for that, and indeed other movies have had scenes with more geometry and textures. The reason was cost and time. It takes a lot of work to do an individual design, so only the main characters got it. The extras were done more simply to keep cost down.

Same shit with games. Even if new tech makes it so that every single brick in a wall in a game could be unique, it probably won't because that costs too much compared to creating just a few brick textures and layering them together along with some detail textures.
 
While Microsoft is hyping this for XBSX, AFAIK this is also in RTX and RDNA2 cards so the HW is likely in Sony PS5 RDNA2 GPU, which means Sony could do something similar.
 
sony most likely could do something similar, so sonys will be even faster if the raw hw is faster
 
Back
Top