ABS123

Since XNA only supports DirectX 9.0

Where is Managed Direct3D 10

I hope there will be Managed wrapper for Direct3D 10

thanks



Re: Direct3D 10 Managed Direct3D 10

Ralf Kornmann

There are signs that after the unmanaged version of the Direct3D 10 API seems stable now Microsoft has start working on the managed one. I am sure we will see a beta in a future DirectX SDK.

Anyway as I wanted a managed wrapper before I have written my own. There is an older version over at MDXInfo. But as it is not compatible with RC1 and the August SDK it will hopefully updated in the next few days with a more complete version.






Re: Direct3D 10 Managed Direct3D 10

ABS123

could anyone from Direct3D 10 team answer my question.

Is there any plan at all for Managed direct3d 10

Or Direct3D 10 will be out of reach for managed code.

I know third parties can wrap Direct3D 10 but it will be far inferior from what Microsoft could offer in term of design, documentation, and performance.

I know at this point of time there is not many games written in managed code, but if you guys(I mean DX team) waiting for that to happen before offering any solutions. It will be highly unreasonable in your part. You must offer a solution first then games will be written in managed code.

The reason I am so concerned is there is so much time being wasted to code in aging C++ language. It is way over due to have a replacement for overly overloaded C++ language.





Re: Direct3D 10 Managed Direct3D 10

Ralf Kornmann

Well, a managed version of DirectX 9 was offered as part of the DirectX 9 beta some years ago. Therefore we have a managed solution for the same time and there are still not many games written in managed code.

I don¡¯t want to sound insulted but what makes you think that Microsoft could build a better designed, better documented and faster managed Direct3D 10 than anybody else

There are rules how to design managed libraries and MDX 1.1 ignore them more than one time for speed reason.

Have you looked at the MDX 1.1 documentation lately

I don¡¯t want to diminish Tom Millers great work but as MDX 1.1 was mainly a one man show they don¡¯t even had a man power advantage. Maybe more resources will be assigned to the managed Direct3D 10 project.

As Microsoft uses the same compiler internal that they give anybody else they have no advantage here. They could only produced a faster managed Direct3D 10 then the rest of us if they ignore the unmanaged Direct3D 10 runtime completely and write a new one in managed code. But with the right documentation anybody else can do this do.

The advantage of an official managed Direct3D 10 version would be that it is official. People will start to write articles and books about it and therefore knowledge will spread faster.






Re: Direct3D 10 Managed Direct3D 10

Michael Oneppo

You¡¯re right in saying that there isn¡¯t that much we could do that you can¡¯t ¨C but we are interested in doing an official version of MDX10 and we¡¯re currently researching our options. What we¡¯re looking to do is have parity with the native APIs as much as possible to ensure developers have the same power in .NET as they do in native. Stay tuned for updates, betas, news¡­

It¡¯s important to note that unless you want D3D10 specific features (for which there¡¯s no hardware available yet) XNA Express is an excellent option for game development. The amount of stuff they provide you to help you get started is amazing and is worth checking out.






Re: Direct3D 10 Managed Direct3D 10

Ralf Kornmann

Michael, it¡¯s good to have an official word about managed Direct3D 10. As I have written before something with official support from Microsoft will always better than my own one man show. But as long as you don¡¯t write a fully managed runtime that talk direct to the driver (I thought about this too) I believe you couldn¡¯t be much faster than my wrapper. Maybe if you have access to a better C++/CLR compiler that allows you to use less expensive managed to unmanaged calls.

If the last rumors are true the D3D10 hardware problem should be solved in the next few days. But I am sure you already know this better but could not talk about it.






Re: Direct3D 10 Managed Direct3D 10

ABS123

Thanks Michael,

That was the answer I was looking for. (Official MDX10 way cool).

I already did check out XNA it is awesome,

But XNA is partially crippled because it had to make so many compromises in order to be multiplatform. That is why pure PC managed DX API is imperative in order for PC to remain competitive gaming platform.

When creating MDX10

1- making close mapping between DX10 and MDX10 will considerably ease up migration from C++ to C#.

2- following .net design pattern make it much easier for .net programmer to use MDX10 since they are already familiar with .net design pattern.

It make sense to maintain a balance between the above design gaols.

You may face a tough fight from .net people. They will demand that MDX10 must adhere to .net design pattern, which is good thing.

I heard CLR team trying to optimize floating and memory management so managed code will be better suited for floating-point intensive application (games).





Re: Direct3D 10 Managed Direct3D 10

Jononet

XNA only lets you use an entire window for the graphicscomponent. This is not much use to people like me that make business apps with 3D elements in them, for us being able to create a device in a control within a Windows Form is essential.

Our applications are never going to hit the XBox so we would end up moving away to OpenGL libraries rather than use C++ which is too much of an overhead in a RAD environment.





Re: Direct3D 10 Managed Direct3D 10

Flavious

I think it would be a good idea for Microsoft to lay out a definitive road-map for the future of Managed DX, so that developers can have a sense of certainty that investments in development time along these lines are not a wasted effort. Hand-waving and deferrals to future SDK releases don't exactly inspire confidence. 

Anyway, having ported much of my native C++ engine to MDX, I can say with confidence that MDX is definitely up to the challenge, as long as you do vector intensive graphics related stuff on the GPU (e.g., skinning), and minimize the use of P/invoke. What I'm really missing is a way to directly tap SSE/3DNOW! from the .NET languages, for physics; perhaps this could be exposed in some sort of "unsafe" context. I've also wondered if specialized hardware is in the future of GC implementations.

Make GC wicked fast, eliminate/reduce P/Invoke overhead, and expose SSE/3DNOW!. That's all, baby! 

 

 

 

 

 





Re: Direct3D 10 Managed Direct3D 10

Rim van Wersch

 Flavious wrote:
I think it would be a good idea for Microsoft to lay out a definitive road-map for the future of Managed DX, so that developers can have a sense of certainty that investments in development time along these lines are not a wasted effort.

Couldn't agree more






Re: Direct3D 10 Managed Direct3D 10

C#Re_eYe

Me too.

I wan't to sense that Microsoft will support us while i am improving my managed game engine.
That is what i expect from the Microsoft i know...




Re: Direct3D 10 Managed Direct3D 10

Flavious

Yes, it's beginning to appear as though it's "every man for himself", where MDX is concerned.

This really seems weird to me, that such a huge, significant part of Windows is being side stepped in .NET.





Re: Direct3D 10 Managed Direct3D 10

zmanvortex

I am extremely frustrated because I have to put my project on hold until Microsoft lets me know where MDX is going.

I have already investigated managed OpenGL wrappers and have come to the following descision.

IF MICROSOFT FORCES ME TO USE XNA FOR NON-GAME APPS THEN I WILL SWITCH TO OPENGL!!!!!!!





Re: Direct3D 10 Managed Direct3D 10

magicalclick

Well, I think it is pretty clear that MS takes MDX seriously. They are moving game developement toward MDX using XNA. Eventually MDX will become the main focus. Even though they didn't say it out loud, they are doing it.

It is just that they have to focus on DX10 right now. If they can't make a good DX10, what's the point to make MDX for DX10, it will be junk anyway

But it will cool if they can release MDX along with DX in the future, or at least in a 6 months time frame.





Re: Direct3D 10 Managed Direct3D 10

Joel Martinez

 Jononet wrote:

XNA only lets you use an entire window for the graphicscomponent. This is not much use to people like me that make business apps with 3D elements in them, for us being able to create a device in a control within a Windows Form is essential.

Our applications are never going to hit the XBox so we would end up moving away to OpenGL libraries rather than use C++ which is too much of an overhead in a RAD environment.

Though it's not stated in the docs, XNA does in fact let you manually create the device and render to a control in a winform. I have an example of doing just that here:
http://codecube.net/item.asp cc_ItemID=338

:-)