Digital Foundry Article Technical Discussion [2023]

Status
Not open for further replies.
So, Starfield is basically broken on any GPU that is not AMD. As AMD GPUs are 40% faster than comparable Intel and NVIDIA GPUs. Hell the game couldn't even boot on Intel Arc GPUs in the beginning, with some Bethesda rep saying Arc GPUs are not supported!

 
R&C doesn't seem to do any special pre-caching to reduce the dependency on IO. It just uses standard Windows file caching.



Direct Storage is primarily about making the IO API itself work more efficiently with high speed SSD's. The previous Win32 IO API was designed around much slower HDD's and thus puts a large overhead on the CPU when dealing with the high throughputs of moderns NVMe's. On the PC side it also has a GPU decompression component which is dealt with by the hardware decompressor on the Xbox. But it's still critical to the efficiency of IO in termsof it's CPU overhead on the Xbox, regardless of the HW decompression.



Strictly speaking there is -AMD's Smart Access Storage. It just requires specific game integration and only works on modern AMD only platforms. And no game has implemented it yet.



No it doesn't. DMA controllers, 'IO co-processors', on chip RAM... these are all just standard components on any decent SSD controller. Sony is just really good at taking standard components, and marketing them as if they are some sort of unique Sony innovation. See: PS4 Move Engines.
R&C doesn't seem to do any special pre-caching to reduce the dependency on IO. It just uses standard Windows file caching ... i recall nixxes claiming they did precaching if direct storage wasnt implemented and digital foundry testing this..

Direct Storage is primarily about making the IO API itself work more efficiently with high speed SSD's. The previous Win32 IO API was designed around much slower HDD's
Yes on pc this is important but on consoles whats the benefit here, hasnt xbox io always had low level access since its a console and doesnt require bloated windows io api's hdd or not.

No it doesn't. DMA controllers, 'IO co-processors', on chip RAM... these are all just standard components on any decent SSD controller. Sony is just really good at taking standard components, and marketing them as if they are some sort of unique Sony innovation. See: PS4 Move Engines.

xbox ssd has no ram in it, ps5 has been found to have 512mb memory on its ssd and sram, cache scrubbers, 12 channel interface vs 4 on xbox. heres what i found reading online and it still corresponds to what i know of both systems there is more hardware on ps5 and less on xbox..

IO Coprocessor​

The IO complex in the PS5's SoC also includes a dual-core processor with its own pool of SRAM. Sony has said almost nothing about the internals of this: Mark Cerny describes one core as dedicated to SSD IO, allowing games to "bypass traditional file IO", while the other core is described simply as helping with "memory mapping". For more detail, we have to turn to a patent Sony filed years ago, and hope it reflects what's actually in the PS5.

The IO coprocessor described in Sony's patent offloads portions of what would normally be the operating system's storage drivers. One of its most important duties is to translate between various address spaces. When the game requests a certain range of bytes from one of its files, the game is looking for the uncompressed data. The IO coprocessor figures out which chunks of compressed data are needed and sends NVMe read commands to the SSD. Once the SSD has returned the data, the IO coprocessor sets up the decompression unit to process that data, and the DMA engine to deliver it to the requested locations in the game's memory.

Since the IO coprocessor's two cores are each much less powerful than a Zen 2 CPU core, they cannot be in charge of all interaction with the SSD. The coprocessor handles the most common cases of reading data, and the system falls back to the OS running on the Zen 2 cores for the rest. The coprocessor's SRAM isn't used to buffer the vast amounts of game data flowing through the IO complex; instead this memory holds the various lookup tables used by the IO coprocessor. In this respect, it is similar to an SSD controller with a pool of RAM for its mapping tables, but the job of the IO coprocessor is completely different from what an SSD controller does. This is why it will be useful even with aftermarket third-party SSDs.

Cache Coherency​

The last somewhat storage-related hardware feature Sony has disclosed is a set of cache coherency engines. The CPU and GPU on the PS5 SoC share the same 16 GB of RAM, which eliminates the step of copying assets from main RAM to VRAM after they're loaded from the SSD and decompressed. But to get the most benefit from the shared pool of memory, the hardware has to ensure cache coherency not just between the several CPU cores, but also with the GPU's various caches. That's all normal for an APU, but what's novel with the PS5 is that the IO complex also participates. When new graphics assets are loaded into memory through the IO complex and overwrite older assets, it sends cache invalidation signals to any relevant caches—to discard only the stale data, rather than flush the entire GPU caches.

What about the Xbox Series X?​

There's a lot of information above about the Playstation 5's custom IO complex, and it's natural to wonder whether the Xbox Series X will have similar capabilities or if it's limited to just the decompression hardware. Microsoft has lumped the storage-related technologies in the new Xbox under the heading of "Xbox Velocity Architecture":

Microsoft defines this as having four components: the SSD itself, the compression engine, a new software API for accessing storage (more on this later), and a hardware feature called Sampler Feedback Streaming. That last one is only distantly related to storage; it's 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.

Since Microsoft doesn't mention anything like the other PS5 IO complex features, it's reasonable to assume the Xbox Series X doesn't have those capabilities and its IO is largely managed by the CPU cores. But I wouldn't be too surprised to find out the Series X has a comparable DMA engine, because that's kind of feature has historically shown up in many console architectures.
source https://www.anandtech.com/show/15848/storage-matters-xbox-ps5-new-era-of-gaming/2

See: PS4 Move Engines
I would argue the opposite is true remember tiled resources, mega textures and now sampler feedback streaming, microsoft are well known for using big blast processing type of pr rhetoric when advertising common technologies they even claimed series x/s ssd is equivalent to 40gb virtual ram and that series s low ram isnt a problem because sfs will increase memory.. ill trust mark cerny any day than any microsoft representative since ive been lied since xbox 360 project milo days by uncle molyneux where as mark cerny backs his words..
 
Last edited:
So, Starfield is basically broken on any GPU that is not AMD. As AMD GPUs are 40% faster than comparable Intel and NVIDIA GPUs. Hell the game couldn't even boot on Intel Arc GPUs in the beginning, with some Bethesda rep saying Arc GPUs are not supported!

Im starting to think the excuse of 30fps wasnt some technical demanding issue other than their engine being completely ancient this explains the consistent loading and bad performance i earlier thought its a cpu issue but ive changed my mind.
 
R&C doesn't seem to do any special pre-caching to reduce the dependency on IO. It just uses standard Windows file caching ... i recall nixxes claiming they did precaching if direct storage wasnt implemented and digital foundry testing this..

The game uses Windows file caching. So like basically everything else, when the data is accessed for the first time, it's access directly from disk. But that process results in it being cached in memory and subsequent accesses to that same data will come from the cache in memory rather than disk - until the cache is overwritten by newer data. This has been tested and proven on this forum.

Yes on pc this is important but on consoles whats the benefit here, hasnt xbox io always had low level access since its a console and doesnt require bloated windows io api's hdd or not.

Xbox runs windows and previous xbox's which used slow HDD's had no reason to use any IO API other than the legacy Windows one because that works perfectly fine for slower drives. Direct Storage was required for Xbox as much as PC because the legacy API is inefficient with high speed SSD's.

Sony have talked about developing their own custom API and firmware as well to run it's own IO system. It's highly unlikely this is the same IO API they used on PS4.

xbox ssd has no ram in it, ps5 has been found to have 512mb memory on its ssd and sram, cache scrubbers, 12 channel interface vs 4 on xbox. heres what i found reading online and it still corresponds to what i know of both systems there is more hardware on ps5 and less on xbox..

IO Coprocessor​

The IO complex in the PS5's SoC also includes a dual-core processor with its own pool of SRAM. Sony has said almost nothing about the internals of this: Mark Cerny describes one core as dedicated to SSD IO, allowing games to "bypass traditional file IO", while the other core is described simply as helping with "memory mapping". For more detail, we have to turn to a patent Sony filed years ago, and hope it reflects what's actually in the PS5.

The IO coprocessor described in Sony's patent offloads portions of what would normally be the operating system's storage drivers. One of its most important duties is to translate between various address spaces. When the game requests a certain range of bytes from one of its files, the game is looking for the uncompressed data. The IO coprocessor figures out which chunks of compressed data are needed and sends NVMe read commands to the SSD. Once the SSD has returned the data, the IO coprocessor sets up the decompression unit to process that data, and the DMA engine to deliver it to the requested locations in the game's memory.

Since the IO coprocessor's two cores are each much less powerful than a Zen 2 CPU core, they cannot be in charge of all interaction with the SSD. The coprocessor handles the most common cases of reading data, and the system falls back to the OS running on the Zen 2 cores for the rest. The coprocessor's SRAM isn't used to buffer the vast amounts of game data flowing through the IO complex; instead this memory holds the various lookup tables used by the IO coprocessor. In this respect, it is similar to an SSD controller with a pool of RAM for its mapping tables, but the job of the IO coprocessor is completely different from what an SSD controller does. This is why it will be useful even with aftermarket third-party SSDs.

Cache Coherency​

The last somewhat storage-related hardware feature Sony has disclosed is a set of cache coherency engines. The CPU and GPU on the PS5 SoC share the same 16 GB of RAM, which eliminates the step of copying assets from main RAM to VRAM after they're loaded from the SSD and decompressed. But to get the most benefit from the shared pool of memory, the hardware has to ensure cache coherency not just between the several CPU cores, but also with the GPU's various caches. That's all normal for an APU, but what's novel with the PS5 is that the IO complex also participates. When new graphics assets are loaded into memory through the IO complex and overwrite older assets, it sends cache invalidation signals to any relevant caches—to discard only the stale data, rather than flush the entire GPU caches.

What about the Xbox Series X?​

There's a lot of information above about the Playstation 5's custom IO complex, and it's natural to wonder whether the Xbox Series X will have similar capabilities or if it's limited to just the decompression hardware. Microsoft has lumped the storage-related technologies in the new Xbox under the heading of "Xbox Velocity Architecture":

Microsoft defines this as having four components: the SSD itself, the compression engine, a new software API for accessing storage (more on this later), and a hardware feature called Sampler Feedback Streaming. That last one is only distantly related to storage; it's 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.

Since Microsoft doesn't mention anything like the other PS5 IO complex features, it's reasonable to assume the Xbox Series X doesn't have those capabilities and its IO is largely managed by the CPU cores. But I wouldn't be too surprised to find out the Series X has a comparable DMA engine, because that's kind of feature has historically shown up in many console architectures.
source https://www.anandtech.com/show/15848/storage-matters-xbox-ps5-new-era-of-gaming/2

As noted above, the IO hardware itself in the PS5 is largely nothing unique. That 512MB memory isn't for IO but rather for running the OS as I understand it (although it's true the XBox does not have an equivalent pool as far as I know). As to the various co-processors and cache, it's nothing special. See here an explanation for what composes a modern Phison controller for example. Note the mention of 20 processor cores (as well as dedicated cache) for handling various IO related tasks. Even the article you quoted above notes that standard SSD's already have processors and cache to manage their own mapping tables.


Where the PS5 is unique is in the jobs is assigns to that hardware, primarily because it differs to a normal PC with it's UMA (which requires a greater degree of cache coherency) and it's hardware decompressor.

So essentially the PS5 has a fairly standard hardware set (outside of the hardware decompressor and cache scrubbers) but a highly customised firmware running it all. Grated the SSD controler in the Xbox is understood to be a bit more basic than that in the PS5 (and higher end PC drives), but it also requires less resources because it's slower overall and thus demands less processing grunt to reach it's full speed potential.

See this page and particularly the post at the bottom by Fabian Giesen (developer of Kraken who's worked very closely with the PS5's IO architecure):


Fabian Giesen said:
Along the same lines, 2 helper processors in an IO block that has both a full Flash controller and the decompression/memory mapping/etc. units is not by itself remarkable. Every SSD controller has one. That's what processes the SATA/NVMe commands, does the wear leveling, bad block remapping and so forth. The special part is not that these processors exist, but rather that they run custom firmware that implements a protocol and feature set quite different from what you would get in an off-the-shelf SSD.

I would argue the opposite is true remember tiled resources, mega textures and now sampler feedback streaming, microsoft are well known for using big blast processing type of pr rhetoric when advertising common technologies they even claimed series x/s ssd is equivalent to 40gb virtual ram and that series s low ram isnt a problem because sfs will increase memory.. ill trust mark cerny any day than any microsoft representative since ive been lied since xbox 360 project milo days by uncle molyneux where as mark cerny backs his words..

My arguing that Sony markets standard features as something unique to their hardware does not preclude the possibility of Microsoft doing the same. That said, I don't recall Microsoft ever stating that tiles resources tiles resources or mega texturing were unique to their consoles. While SFS itself does actually feature a unique hardware element in the series console for blending mips together when the higher res mip fails to load in time for the fame. It is true though that Sampler Feedback streaming (little s) is also possible on any DX12U level hardware, just without that hardware based blending element. The xbox USP there has likely been significantly overblown.
 
Why do ugly ganes like starfield run bad and gorgeous games like plague tale run well.. developer talent and engines technology make the difference 😁😉💻🎮
 
Both Starfield and Plague Tale Requiem run 1440p reconstructed to 4k at 30fps on XSX and 900p upped to 1440p at 30fps on XSS, so equally bad/well.
 
Both Starfield and Plague Tale Requiem run 1440p reconstructed to 4k at 30fps on XSX and 900p upped to 1440p at 30fps on XSS, so equally bad/well.
What about the PC version?

Not that it matters because I wouldn’t really compare these games anyway.
 
The game uses Windows file caching. So like basically everything else, when the data is accessed for the first time, it's access directly from disk. But that process results in it being cached in memory and subsequent accesses to that same data will come from the cache in memory rather than disk - until the cache is overwritten by newer data. This has been tested and proven on this forum.



Xbox runs windows and previous xbox's which used slow HDD's had no reason to use any IO API other than the legacy Windows one because that works perfectly fine for slower drives. Direct Storage was required for Xbox as much as PC because the legacy API is inefficient with high speed SSD's.

Sony have talked about developing their own custom API and firmware as well to run it's own IO system. It's highly unlikely this is the same IO API they used on PS4.



As noted above, the IO hardware itself in the PS5 is largely nothing unique. That 512MB memory isn't for IO but rather for running the OS as I understand it (although it's true the XBox does not have an equivalent pool as far as I know). As to the various co-processors and cache, it's nothing special. See here an explanation for what composes a modern Phison controller for example. Note the mention of 20 processor cores (as well as dedicated cache) for handling various IO related tasks. Even the article you quoted above notes that standard SSD's already have processors and cache to manage their own mapping tables.


Where the PS5 is unique is in the jobs is assigns to that hardware, primarily because it differs to a normal PC with it's UMA (which requires a greater degree of cache coherency) and it's hardware decompressor.

So essentially the PS5 has a fairly standard hardware set (outside of the hardware decompressor and cache scrubbers) but a highly customised firmware running it all. Grated the SSD controler in the Xbox is understood to be a bit more basic than that in the PS5 (and higher end PC drives), but it also requires less resources because it's slower overall and thus demands less processing grunt to reach it's full speed potential.

See this page and particularly the post at the bottom by Fabian Giesen (developer of Kraken who's worked very closely with the PS5's IO architecure):






My arguing that Sony markets standard features as something unique to their hardware does not preclude the possibility of Microsoft doing the same. That said, I don't recall Microsoft ever stating that tiles resources tiles resources or mega texturing were unique to their consoles. While SFS itself does actually feature a unique hardware element in the series console for blending mips together when the higher res mip fails to load in time for the fame. It is true though that Sampler Feedback streaming (little s) is also possible on any DX12U level hardware, just without that hardware based blending element. The xbox USP there has likely been significantly overblown.
see here an explanation for what composes a modern Phison controller for example. Note the mention of 20 processor cores (as well as dedicated cache) for handling various IO related tasks. Even the article you quoted above notes that standard SSD's already have processors and cache to manage their own mapping tables.

Im aware of this i wasnt generally comparing ps5 ssd to pc ssds im comparing what the overall io hardware is on ps5 vs the series x as someone here claimed they where interchangeable..

And its still not known if the 512mb ram on ps5 is for os this was just early speculation but mark cerny specifically said the io block uses a big block of sram so all this plus cache scrubbers 12 channel interface already make it advanced than the xbox solution and i remember cerny's goal wasnt just shortening loading times he wanted no loading screens but microsoft didnt take such an approach.

Microsoft ever stating that tiles resources tiles resources or mega texturing were unique to their consoles
i remember a dozen years ago during xbone unveal tiled resources had a big pr campaign from microsoft promising the same things virtual memory fast forward 2020s they still talk about virtual texture tricks as ram extensions and people fall for it now 3 years on and devs are constantly complaining about series s while consumers are criticizing developers for not using sfs as they where fooled by ms.

My arguing that Sony markets standard features as something unique
Cerny nor sony didnt market ps5 solution as unique ever... people marketed it for themselves after the road to ps5 and at the time there was no 5gb/s ssd's with 8-22g/s decompression speeds such ssd's came months later, ps3 cell was also unique, all companies lie but sony has a history of backing its words where as microsoft record has doubts from project milo to cloud physics simulations, sfs memory saver and 40gb virtual memory... if i had a penny everytime microsoft lied id have a bank by now.
 
Cerny nor sony didnt market ps5 solution as unique ever... people marketed it for themselves

Well... yes... like you did on the previous page when you said:

kraken and bcpack decompression aside ps5 has alot of io hardware not present on xbox or pc its not interchangeable really,

And you referenced the Sony presentation in making this assumption which has clearly led to you to believe that the mentioned hardware there is unique. The point I've been making since then is that it's not, at least not in relation to the PC.

In fact the IO hardware (aside from the decompression block) in the PS5 is roughly equivalent to a lower high end PCIe 4 PC SSD while the the Xbox is more equivalent to a mid range unit. The Xbox still has "helper cores" but only 1 vs the 2 in PS5. It also has access to the dedicated SRAM cache just like the PS5:


Incidentally, the number of memory channels in itself isn't a specific advantage and needs to be considered alongside the speed of flash memory used. Sure the PS5 has 12 channels vs the Xbox's 4 (and even 8 in top end PCIe5 PC drives) but it simultaneously uses slower memory (probably to save on BoM) which means the net result is SSD throughput iwhich is around 2.3x faster rather than the 3x that the memory channels would imply.

after the road to ps5 and at the time there was no 5gb/s ssd's with 8-22g/s decompression speeds such ssd's came months later,

I'm not sure what point you're trying to make here. Simply announcing a product is not the same as it being available to the market. There may have been no PCIe4 SSD's in the market when the PS5 drive was announced, but the PCIe4 roadmap had already been known for a long time before this. And by the time the PS5 actually released, it's drive speed was not unique (although other elements of the IO architecture were).

ps3 cell was also unique

No-one argued that Sony doesn't make unique hardware, or course they have. But not everything that is marketed is such a way as to suggest it is unique, actually is unique.
 
Well... yes... like you did on the previous page when you said:



And you referenced the Sony presentation in making this assumption which has clearly led to you to believe that the mentioned hardware there is unique. The point I've been making since then is that it's not, at least not in relation to the PC.

In fact the IO hardware (aside from the decompression block) in the PS5 is roughly equivalent to a lower high end PCIe 4 PC SSD while the the Xbox is more equivalent to a mid range unit. The Xbox still has "helper cores" but only 1 vs the 2 in PS5. It also has access to the dedicated SRAM cache just like the PS5:


Incidentally, the number of memory channels in itself isn't a specific advantage and needs to be considered alongside the speed of flash memory used. Sure the PS5 has 12 channels vs the Xbox's 4 (and even 8 in top end PCIe5 PC drives) but it simultaneously uses slower memory (probably to save on BoM) which means the net result is SSD throughput iwhich is around 2.3x faster rather than the 3x that the memory channels would imply.



I'm not sure what point you're trying to make here. Simply announcing a product is not the same as it being available to the market. There may have been no PCIe4 SSD's in the market when the PS5 drive was announced, but the PCIe4 roadmap had already been known for a long time before this. And by the time the PS5 actually released, it's drive speed was not unique (although other elements of the IO architecture were).

No-one argued that Sony doesn't make unique hardware, or course they have. But not everything that is marketed is such a way as to suggest it is unique, actually is unique.
the overall argument was if ps5 io is interchangeable to the series x one and youve proved my point, it isnt
this is from the article you referenced..

Instead of using raw hardware, the Series X/S consoles will instead utilize a combination of storage-specific APIs and software toolsets from the Velocity Architecture stack. This includes DirectStorage API from DX12 Ultimate that massively streamlines data flows, as well as new features like Sampler Feedback Streaming and more.

Read more: https://www.tweaktown.com/news/7572...-nand-phison-e19-memory-controller/index.html

And you referenced the Sony presentation in making this assumption which has clearly led to you to believe that the mentioned hardware there is unique. The point I've been making since then is that it's not, at least not in relation to the PC.

again it didnt just led me to believe that its unique, developers and fans alike said the same thing, not only that from comparison data that we all saw ps5 games have always loaded faster sometimes even faster than 7gbs pc ssds from spider man to ratchet and other games alike so the bottlenecks are still present even with direct storage on pc, seems it wasnt just marketing.

I'm not sure what point you're trying to make here. Simply announcing a product is not the same as it being available to the market. There may have been no PCIe4 SSD's in the market when the PS5 drive was announced, but the PCIe4 roadmap had already been known for a long time before this. And by the time the PS5 actually released, it's drive speed was not unique (although other elements of the IO architecture were).

mark cerny said the same thing when talking about ps5 compatible drives, he said within a year bespoke drives will come out that will be compatible with ps5 but i remember very well even when ps5 came out such drives werent out yet.. until for a couple of months.. so again it wasnt just pr.
 
the overall argument was if ps5 io is interchangeable to the series x one and youve proved my point, it isnt
this is from the article you referenced..

Instead of using raw hardware, the Series X/S consoles will instead utilize a combination of storage-specific APIs and software toolsets from the Velocity Architecture stack. This includes DirectStorage API from DX12 Ultimate that massively streamlines data flows, as well as new features like Sampler Feedback Streaming and more.

Read more: https://www.tweaktown.com/news/7572...-nand-phison-e19-memory-controller/index.html

That quote is taken completely out of context. The author is referring to the raw throughput of the PS5 drive (5.5GB/s) which we know is faster than the Xbox's 2.4GB/s. The context of the above statement is that the Xbox will mitigate it's throughput disadvantage via software solutions like SFS. I don't agree with the statement, but that's the correct context.

And I believe @iroboto's original statement that they are interchangeable was referring to the general architecture, not the actual throughput.

again it didnt just led me to believe that its unique, developers and fans alike said the same thing, not only that from comparison data that we all saw ps5 games have always loaded faster sometimes even faster than 7gbs pc ssds from spider man to ratchet and other games alike so the bottlenecks are still present even with direct storage on pc, seems it wasnt just marketing.

Some games load faster on the PS5, others load faster on PC (e.g. Forspoken). And in those instances where the game loads faster on PS5, I don't believe it has ever been shown to be down to IO throughput. In R&C for example is doesn;t matter if you have a 3.5GB/s drive or a 10GB/s drive, the loading times are the same.

CPU setup time is going to be the bottleneck in the overwhelming majority of cases, and here the PS5 can leverage a few console specific advantages (which would also apply to the Xbox) such as the ability to load BVH structures from disk rather than calculate them at run time, the lack of need to do any shader compilation, and of course, the ability to offload decompression work to a dedicated chip (which PC's now have an answer to in the form of Direct Storage GPU decompression.

mark cerny said the same thing when talking about ps5 compatible drives, he said within a year bespoke drives will come out that will be compatible with ps5 but i remember very well even when ps5 came out such drives werent out yet.. until for a couple of months.. so again it wasnt just pr.

No, there were 7GB/s drives available in the PC space before the PS5 launched.

 
So AMD was apparently helping devs get the most out of FSR with Starfield and they...forget to account for the mip bias issue.

Jesus.

Oh, and:

reddit user Nefsen402 said:
Vkd3d (the dx12->vulkan translation layer) developer has put up a change log for a new version that is about to be (released here) and also a pull request with more information about what he discovered about all the awful things that starfield is doing to GPU drivers (here).

Basically:

  1. Starfield allocates its memory incorrectly where it doesn't align to the CPU page size. If your GPU drivers are not robust against this, your game is going to crash at random times.
  2. Starfield abuses a dx12 feature called ExecuteIndirect. One of the things that this wants is some hints from the game so that the graphics driver knows what to expect. Since Starfield sends in bogus hints, the graphics drivers get caught off gaurd trying to process the data and end up making bubbles in the command queue. These bubbles mean the GPU has to stop what it's doing, double check the assumptions it made about the indirect execute and start over again.
  3. Starfield creates multiple `ExecuteIndirect` calls back to back instead of batching them meaning the problem above is compounded multiple times.
What really grinds my gears is the fact that the open source community has figured out and came up with workarounds to try to make this game run better. These workarounds are available to view by the public eye but Bethesda will most likely not care about fixing their broken engine. Instead they double down and claim their game is "optimized" if your hardware is new enough.

😬*

Alex was extremely generous it seems.

Edit: As @Andrew Lauritzen pointed out this may be a red herring, at least in the implication Bethesda is doing something grievously improper here.
 
Last edited:
So AMD was apparently helping devs get the most out of FSR with Starfield and they...forget to account for the mip bias issue.

Jesus.

Oh, and:



😬

Alex was extremely generous it seems.

I find it super weird that negative mip map bias gets forgotten about in games with upscaling. I remember some of the early games it was an issue, but maybe it wasn't explicitly a recommendations back then. To miss it now is just a big oversight.

Don't know enough to comment about the renderer. It sounds like ExecuteIndirect is dispatching compute shaders, but some of them don't actually lead to draws, and they're not batched which also causes poor utilization because of "bubbles" or bottlenecks. Is this a barrier issue, where you end up with more barriers as sync points because each executeindirect requires its own barrier? I've only superficially read about this stuff.
 
There were some id software people that worked on the game. Not like there isn't the talent there to do it. There's probably some reason why it's not an easy fix. Unless they do a siggraph post-morten or something, will probably be hard to find out what the deal is.
 
As to why PS5 performs so well, it is a lead platform to develop for. We have seen many titles on Series consoles that after launch continue to improve stuttering performance to be eventually as smooth as ps5. And secondly stuttering is not likely a result of asset loading challenges, but instead API call problems that need additional sorting.
nope - PS5 is much much better engineered system - thats why it is performing better. And you cannot seriously state here that all that PS5 brings is old news. LOL
Hm PS5 seems particular in this forum being hated and its tech marginalised. Wich in itself is not that big of a deal but it is suprising given the "standard" the forum sets for itsself to be "pro tech" . PS5 when it came out was ahead of everything else and still is in many areas of tech. It simply features the most elegant way of Power! With PS5 we saw the first time AAA games with next to none loading screens. If everything was old tech and nothing special where are your AAA Games between lets say 2015 to 2020 that featured "no load times" like PS5 does show it since release?
Where are they?
It is so ridiculous.

ah i see a certain member dwells again in his anti PS5 Propaganda. Time for the next wave of PS5 Exclusives that show that what PS5 can achieve is not so easy to mimic on PC if the console is used to its fullest ;)
 
Last edited:
nope - PS5 is much much better engineered system - thats why it is performing better. And you cannot seriously state here that all that PS5 brings is old news. LOL
Hm PS5 seems particular in this forum being hated and its tech marginalised. Wich in itself is not that big of a deal but it is suprising given the "standard" the forum sets for itsself to be "pro tech" . PS5 when it came out was ahead of everything else and still is in many areas of tech. It simply features the most elegant way of Power! With PS5 we saw the first time AAA games with next to none loading screens. If everything was old tech and nothing special where are your AAA Games between lets say 2015 to 2020 that featured "no load times" like PS5 does show it since release?
Where are they?
It is so ridiculous.

ah i see a certain member dwells again in his anti PS5 Propaganda. Time for the next wave of PS5 Exclusives that show that what PS5 can achieve is not so easy to mimic on PC if the console is used to its fullest ;)
Yea, you need to go.

The WAY OF POWER with PS5!!

Screenshot-2023-09-10-164326.png
 
Status
Not open for further replies.
Back
Top