I take it your expertise in software design and implementation is extensive. Sorry, could not resist (pun intended).
Newell: Win8 is a catastrophe; Pardo: I don't disagree.
- Thread starter Grall
- Start date
- Status
I take it your expertise in software design and implementation is extensive. Sorry, could not resist (pun intended).
CouldntResist
Regular
Name one Direct3D 10 feature that you think is impossible/difficult to implement in XP, and example of game/application that depends on it.
The API to channel the hardware-abstracted information is rigid. The driver-code maybe wants to offer the feature, and the OS wants to use it, but there is no way those two can come together.
Microsoft doesn't add/plan extension-mechanisms for their low-level APIs, like those for high-level APIs fe. the OLE-layer. The result is that you get "funky" hacks like ATI did with their cool DX9 add-ons. But basically it ends there. We won't see partially-resident textures in DX for the near future, because there is simply no way to add it in any meaningfull way to the old APIs.
Microsoft doesn't add/plan extension-mechanisms for their low-level APIs, like those for high-level APIs fe. the OLE-layer. The result is that you get "funky" hacks like ATI did with their cool DX9 add-ons. But basically it ends there. We won't see partially-resident textures in DX for the near future, because there is simply no way to add it in any meaningfull way to the old APIs.
Dominik D
Regular
It's not about DX API, it's about driver model. Graphics stack changed completely between XP and Vista and shell (UI) as well as IE are tightly coupled with it for performance reasons. If you pull WDDM in, you have to change the underpinnings of Explorer, UI controlls, IE and probably a dozen more applications. The alternative is updating old driver model with new features. How many HW vendors would update their XP drivers with tons of new features? Here's a guess: none would.
CouldntResist
Regular
This is entirely irrelevant, driver model is not a D3D feature.
You can prove me wrong, by giving me example of game/application that "uses Vista driver model".
Of course I know you can't do it, but I needed to illustrate you technical absurdity of this explanation.
What you say provides factual reasons why there was need for D3D 3, D3D 5, D3D 6, D3D 7, D3D 8, D3D 9, D3D 10, and D3D 11. But it has nothing to do with nonexistence of D3D 10 for XP.
Last edited by a moderator:
Blazkowicz
Legend
The main feature of Direct3D 10/11 is running on that new driver model, then you have geometry shader and tesselators as little features we can live without (until the new consoles use them, that is).
How does that help?, well I guess the newer way is less wasteful of CPU cycles.
Hell, I'd like Microsoft to implement Direct2D in Linux so I can try the acceleration in Firefox, I used pirated Windows before so they owe me, right?
How does that help?, well I guess the newer way is less wasteful of CPU cycles.
Hell, I'd like Microsoft to implement Direct2D in Linux so I can try the acceleration in Firefox, I used pirated Windows before so they owe me, right?
Dominik D
Regular
Driver model has everything to do with this. Drivers on Vista+ are built around a completely different infrastructure provided by the OS than they were on XP. All of the drivers had to be mostly written from scratch. The way driver is being notified about stuff application does in DirectX is different; when it's notified changed; how you divide resources have changed; everything is different. DirectX APIs "just work" because there's a DX infrastructure and gfx driver underneath that deal with applications' requests.
You have to realize that you can't get DX10/11 features without support from the driver. You just can't. So what and how driver does matters profoundly. DXGI doesn't exist on XP and it's required to even start any DX10/11 application. And none of the XP drivers supported it because it didn't exist on XP. And you can't just make it part of the XPDM because it requires system-wide facilities for virtual video memory, proper scheduling, fault tolerance and many others. These come with WDDM.
There are only 3 ways to get DX10 features on XP:
- rev DX9.0c to DX9.0d or DX9.1 or something, essentially forking and fragmenting application development even more, requiring hardware manufacturers to work on two generations of drivers in parallel
- port WDDM to XP with everything that comes with it, pretty much making a completely new Windows XP (remember not to charge consumers for this work, after all they are certain it was done magically by itself)
- create a fake DX10/11 layer on top of DX9 making APIs consistent but exposing only DX9-level features through DX10/11 APIs, essentially changing nothing
I simply couldn't resist adding this: go and write some DX9 code, then write some DX10 or DX11 code. Go read XPDM and WDDM documentation. Then come back discussing what's possible and what isn't. And possible doesn't mean "can be done with infinite time and resources", it means "within a reasonable time frame by a reasonably sized team". We occupy this space called reality, stick with it.
CouldntResist
Regular
If this is your way to say you can't name one, I rest my case.
CouldntResist
Regular
Driver model is your red herring. I gave you simple task:
Name one Direct3D 10 feature that you think is impossible/difficult to implement in XP, and example of game/application that depends on it.
That's by far the easiest possible way for you to prove me wrong and that reasons for nonexistence of DX10 on XP were indeed technical. Allegedly there are looooooooooots of features. Vista has been out there for 6 years, so there are lots of apps too. If what you believe were true, the evidence would be at your reach of your hand.
DX10 application cannot depend on something, that isn't part of the DX10 interface and its contract. What's inside the black box, is private matter beteen Windows devs and driver devs. Game/app developer doesn't need nor want to have anything to do with the insides. Unless he is dealing with driver or OS bug, obscure undefined behaviour etc. but that's outside of our scope.
So, instead of producing the evidence, you responded with elaborate mumbo jumbo vomit, that is irrelevant to the task given. Irrelevancy aside, you made some kinda vague claims that need to be addressed:
- You seem to imply that Vista's WDDM is part of the only possible implementation of DX10 interface.
That would be absurd for anyone who can comprehend basic concepts of software engineering.
- You seem to imply that DX10 level hardware cannot function without Vista's WDDM.
For this, there is one giant, well known counter evidence. It had 20th birthday this summer (hint).
Hardware manufacturers do have been working on two generations of drivers in parallell since Vista came out. They continually release Windows XP DX9 drivers for their DX10 and DX11 level hardware. The work hasn't stopped. It's just that one of those two driver generations has to be limited to DX9 level.
For once again, you are making assumption that DX10 level hardware magically requires Vista WDDM to function.
Great, now all you need is to explain exactly why Windows devs would have to refrain from going one step further and exposing also DX10-level features. What magical force would stop them from throwing in support for geometry shaders, texture arrays, SM 4.0?
OpenGL managed to do it with just that, throwing in bunch of ARB extensions. But you believe that Microsoft wouldn't be able to do it with clean rewrite of their own API.
Oh, it's "fake" API, I forgot. Your fake DX10 would be enough for any DX10 game out there to work. Even Wine allows some DX10 games work in entirely alien operating system, but you believe Microsoft wouldn't be able to do it with all games, in its native environment.
You are the one who believes in extraordinary claims without evidence, and against counterevidence.
And save your efforts on this aggressive rhetoric, no amount of it will make you right.
- Status