Digital Foundry Article Technical Discussion [2023]

Status
Not open for further replies.
EDIT: I've also seen nothing to suggest Microsoft is moving to Windows 12. I'm guessing you're talking about some errant article briefly mentioning it, which could have been a typo, misunderstanding or some other mistake most likely. It is not a substantiated rumor, much less any kind of real information to begin with.
Not a typo, nor a misunderstanding.

I guess we'll see.
 
EDIT: I've also seen nothing to suggest Microsoft is moving to Windows 12. I'm guessing you're talking about some errant article briefly mentioning it, which could have been a typo, misunderstanding or some other mistake most likely. It is not a substantiated rumor, much less any kind of real information to begin with.
There are other articles out there. Personally I hope not. I wasn't planning of moving the organization I work for to Windows 11 until 2024 now I might have to deal with a less than stable Windows 12 upgrade.
 
Some important things were left out at the start.
Some theorises AMD kinda strong armed Microsoft into doing DX12 in haste.

The problem facing AMD was their performance in DX11 and OpenGL, they couldn't optimize their drivers to deliver optimal performance in these APIs, and they were severely limited CPU wise, Digital Foundry showed that in many games, NVIDIA GPUs delivered faster performance than their AMD counterparts when paired with low end CPUs in DX11. OpenGL games ran quite horribly on AMD GPUs, and DX11 games ran into performance defeciets too, back then AMD lacked the resources needed to properly optimzie for DX11/OGL like NVIDIA did, they only got to do so recently, when they released their major Adrenaline driver targeting DX11/OGL improvements last year.

So AMD leaned heavy on their advantage in the 8th gen console cycle, they were the provider for both PS4 and Xbox One, so they developed the lower level Mantle proprietary API and encouraged mutli platform developers to port their console codes to that proprietary API, they planned for many developers to support it. AMD sought to alter their rather difficult position with DX11/OGL through Mantle, and these Mantle games performed vastly better with Mantle than DX11 on AMD GPUs, negating AMD's disadvantage.

Microsoft acted rather quickly, probably for fear of Mantle fracturing their API dominance, do they formulated DX12 with backward compatibility with DX11 GPUs, after which AMD ceased all of their Mantle efforts and scrubbed it's support from most of the planned games. Declaring DX12 satisfactory.

However both Mantle and DX12 approaches had major problems, Mantle required more VRAM to run, and required more support from AMD for upcoming new products, as newer GPUs required driver updates from AMD to run the Mantle path well. DX12 inherited some of it's problems too, but added even more.

It was quite clear DX12 was developed in haste though, without giving many of it's features the proper think through, and while the initial wave of DX12 games were satisfactory to AMD and achieved their Mantle goals, AMD quickly got entangled into the barrage of DX12 problems, like everyone else, and in many many cases, the results were not satisfactory anymore.
 
I would not expect a Multiplatform game to use the MS branch of Unreal on PlayStation, but I would expect them to use the MS optimized branch for the MS platforms,

Why would MS have their own branch of Unreal Engine? It is way more effective to just get your fixes into Epic main branch and then don't worry about it.
 
Microsoft acted rather quickly, probably for fear of Mantle fracturing their API dominance, do they formulated DX12 with backward compatibility with DX11 GPUs, after which AMD ceased all of their Mantle efforts and scrubbed it's support from most of the planned games. Declaring DX12 satisfactory.
I thought Vulkan was the main inheritor of Mantle -- what's the history there?
 
Why would MS have their own branch of Unreal Engine? It is way more effective to just get your fixes into Epic main branch and then don't worry about it.

XBOX vs windows, for a start. same GDK backend ( mostly ) but very different optimization strategies.

Maybe MS should release some helper libs that can eaasily be integrated into existing DX12 systems, that do all the memory management for the end user, as a cost of less efficiency,
but easier to use and less complexity overall.

Anyone see Sebbi's twitter posts on a GPU based mem allocater recently?
 
Baldur's Gate 3 is a big enough IP or, should I say, an IP with enough exposure that if Microsoft has not sent a team over to the developer the help them out with this, they should seriously consider doing so. I think the same situation happened with the Matrix Awakenings demo, and the Coalition needed to be called in. I believe it was worth it with that demo. I agree with John's suggestion here.

Though it would cost Microsoft a great deal of money to help developers with the engineering part of their games, games that are not exclusive or have a marketing deal. On a case-by-case bases, Microsoft should consider helping developers get gameplay features they are having trouble with up and running. Only after that fails should Microsoft allow a feature to be dropped on the Series S, IMO.
BG3 is made by the developers of two of my favourite games to date, Divinity Original Sin Enhanced Edition and Divinity Original Sin 2. I am sure the developers are going to deliver handsomely.

Most of my gaming time nowadays is used on coop games to play with my nephews and so on. Some of them classic games. I feel at easy that way, without being always on trend.
 
Baldur's Gate 3 is a big enough IP or, should I say, an IP with enough exposure that if Microsoft has not sent a team over to the developer the help them out with this, they should seriously consider doing so. I think the same situation happened with the Matrix Awakenings demo, and the Coalition needed to be called in. I believe it was worth it with that demo. I agree with John's suggestion here.

Though it would cost Microsoft a great deal of money to help developers with the engineering part of their games, games that are not exclusive or have a marketing deal. On a case-by-case bases, Microsoft should consider helping developers get gameplay features they are having trouble with up and running. Only after that fails should Microsoft allow a feature to be dropped on the Series S, IMO.

So based on what was said, and looking through some older articles, I would agree that VRAM could be the issue here in its current state.

According to statements in this article

“I think especially for Baldur’s Gate 3 – which will have just like, higher [expectations] and bigger attention from everyone – I think we’ll work on it, but let’s first make it for PC. Let’s make it work.”

As we can see, console porting was never discussed even as far back as 2020. So it’s just Stadia and PC at this point in time. Unexpectedly, Stadia dies.

“I don’t think that current-gen consoles would be able to run it. There’s a lot of technical upgrades and updates that we did to our engine, and I don’t know if it would be capable of being able to actually run on those things.

“Maybe it could run, but then we would have to tone down the textures and this and that and it wouldn’t look as cool anymore.”

From this we can gather that they aimed to push their base requirement on PC, maximizing the available memory on the video cards while pushing slightly more on system memory puts series S into a precarious position because it has a sliver more memory than last Gen.

It could be worked around, but it would require help. I think the idea that Series S is holding back Series X is a bit of a modification of truth here. The game was never designed for consoles to begin with, and is now being back ported. I just don’t think they have had sufficient time to work on it, and everyone was upset that this became a PS5 exclusive And developers felt the need the respond.

Remember, stadia died recently, and now we suddenly see it on state of play. I don’t believe in coincidences, they were going to make projections as soon as they lost one of their platforms or financial backing from google.
 
From this we can gather that they aimed to push their base requirement on PC, maximizing the available memory on the video cards while pushing slightly more on system memory puts series S into a precarious position because it has a sliver more memory than last Gen.

I don't know how it runs in practice but going by the system requirements it doesn't seem to be all that aggressive in pushing memory.

Min - 8GB RAM, 4GB VRAM (actually less since it uses a GTX 970)
Rec - 16GB RAM, 8GB VRAM

Min CPU is a FX8350 and Rec. is a R5 3600. Massive HD space requirement at 150GB but no SSD requirement, maybe room for optimization there?
 
I don't know how it runs in practice but going by the system requirements it doesn't seem to be all that aggressive in pushing memory.

Min - 8GB RAM, 4GB VRAM (actually less since it uses a GTX 970)
Rec - 16GB RAM, 8GB VRAM

Min CPU is a FX8350 and Rec. is a R5 3600. Massive HD space requirement at 150GB but no SSD requirement, maybe room for optimization there?
Yea I’ve wondered about that as well. If PC minimal spec can do split screen coop I don’t see why series S can’t. But without more details a bit harder to call this one. It does show that this is a PC first title that has plans to be ported later. Seems like some priorities were shuffled once stadia closed.
 
There are other articles out there. Personally I hope not. I wasn't planning of moving the organization I work for to Windows 11 until 2024 now I might have to deal with a less than stable Windows 12 upgrade.
Fair enough, I missed this.

Seems utterly bizarre, though.

EDIT: Actually, I'm still seeing little evidence or confirmation reading through this. Much seems like speculation, and it's all built on the single line about Meteor Lake supporting Windows 12. That's very light evidence.
 
I don't know how it runs in practice but going by the system requirements it doesn't seem to be all that aggressive in pushing memory.

Min - 8GB RAM, 4GB VRAM (actually less since it uses a GTX 970)
Rec - 16GB RAM, 8GB VRAM

Min CPU is a FX8350 and Rec. is a R5 3600. Massive HD space requirement at 150GB but no SSD requirement, maybe room for optimization there?

Even min spec is still a few GB more than XSS offers.
 
And even then, min specs do not even mean game will run respectably. It could run at sub 30 FPS with super super unstable frametimes for all we know while loading garbo textures. That is simply not desirable or acceptable on S. Maybe game turns everything into N64 textures when you play split coop with 8 gigs of ram and 4 gigs of VRAM. Minimum is minimum. It is literally bare minimums and at times it literally means sub 30 FPS with horrendous graphics. It is not something that you can deploy on Series S. or it would turn into a PR nightmare for Msoft if devs treat it as the "min spec PC". Min spec PCs are min spec PCs, they are treated garbage. They get n64 like textures with horrible reduced quality everywhere, just to get somewhat semi stable frametimes here and there. Even the dreaded Xbox One never stooped so low.
 
I don't know how it runs in practice but going by the system requirements it doesn't seem to be all that aggressive in pushing memory.

Min - 8GB RAM, 4GB VRAM (actually less since it uses a GTX 970)
Rec - 16GB RAM, 8GB VRAM

Min CPU is a FX8350 and Rec. is a R5 3600. Massive HD space requirement at 150GB but no SSD requirement, maybe room for optimization there?
BG3 doesn't seem too VRAM intensive judging by YouTube. This video below seems to place it at ~6GB of VRAM (at 4k with DLAA enabled) and ~5GB for the process. The latter number might be more troublesome but it's hard to say since it's running on Windows.

 
BG3 doesn't seem too VRAM intensive judging by YouTube. This video below seems to place it at ~6GB of VRAM (at 4k with DLAA enabled) and ~5GB for the process. The latter number might be more troublesome but it's hard to say since it's running on Windows.


Series-S offers 8GB total to developers so it very well may be a RAM issue.
 
Custom engines unfortunately come with their own major issues, hence why many devs are actively leaving theirs behind. I don't honestly know if there is a real silver bullet to these issues on PC...but I hope for PC gamers one day this crap gets figured out
hope so... it's not even about consoles vs PC anymore, it's about quality of life and having a bare minimum quality. PC gamers usually focused the advantages of a PC on how powerful they can be. That's wrong, current consoles are more powerful than 90%+ of PCs out there. The advantage of PC gaming is the freedom, thousands of games can be played on a PC, games that could never be published anywhere else. Then the modding community, which I hope to join soon when Return of Rome is released and make campaigns of my favourite RTS ever, AoE1, into AoE2 with its improvements.

It shouldn't be that hard to optimise quite a bit, having a mid tier PC to perform some testing and get realistic results, this would greatly benefit top tier PCs when a game is released.
 
Last edited:
PC gamers usually focused the advantages of a PC on how powerful they can be. That's wrong, current consoles are more powerful than 90%+ of PCs out there.

Based on the Steam Hardware Survey its more like the consoles are as fast or faster than around 80% of PC's while the top 20% of the PC market is as faster or faster than the consoles.

Although it's safe to assume that the vast majority of the PCMR crowd that care about the PCs higher performance sit in that top 20%.
 
hope so... it's not even about consoles vs PC anymore, it's about quality of life and having a bare minimum quality. PC gamers usually focused the advantages of a PC on how powerful they can be. That's wrong, current consoles are more powerful than 90%+ of PCs out there. The advantage of PC gaming is the freedom, thousands of games can be played on a PC, games that could never be published anywhere else. Then the modding community, which I hope to join soon when Return of Rome is released and make campaigns of my favourite RTS ever, AoE1, into AoE2 with its improvements.

It shouldn't be that hard to optimise quite a bit, having a mid tier PC to perform some testing and get realistic results, this would greatly benefit top tier PCs when a game is released.

The thing I love most about PC gaming is that for the most part games are eternal. If I wanted to play a game from 2005 I still can. Console gaming has always felt generational to me. You can only play this game on this specific hardware with this specific controller and there's an expiration date on all of that. The other thing that's really important is the greater freedom to tweak and customize the experience though it seems console games are starting to get decent settings menus too.
 
Status
Not open for further replies.
Back
Top