whoa sick Oo
well i dont think that you could run any game with a 10.584000pixel screen setup because dx10 is limited to 8.192000pixel, so you will be forced to reduce the resolution of your screens to 1440x900
It was my understanding that 8192x8192 for DX10 was the Texture size limit. Not the surface size limit. Am I wrong about this? (I very well could be)
also since I'm on XP with DX9.0c I would be at the 4096x4096 limit which I'm sure 7680x1024 exceeded (assuming the limits are hard for width x height.)
The above said, who are the DirectX experts on the board?
1) Are 4096x4096 and 8192x8192 texture limits or surface limits?
2) Are they hard boundaries for width x height, or maximum memory taken up? ie:
4096x4096=6861881344
7680x1024= 7864320
5040x2100= 10584000
hence why 7680x1024 can be rendered under DX9?
I'm asking the above questions independent of hardware limits (ie what the DX9/10 APIs allow.)
Can you point me at MSDN documentation for the above? I'd like to understand the real limits.
Also if there's anybody who's very familiar with DX9/10 and/or Windows Device drivers (DDK) ??? I've got some ideas for a shim (like softTH but meant for single cards and/or routing calls to specific cards) I'm not really familiar with the DX APIs, but I'd be willing to learn. In a past life I had done some OpenGL work, but it was in the very early openGL days (back on silicon graphics IRIX systems... so it was like OpenGL 1.1 I think.)
If I'm using the term surface wrong in terms of DirectX please let me know, it's my understanding that a surface is a directX "window" that's rendered into [not a surface of a triangle that you are applying a texture to)
3) Can anybody explain how directX surfaces are created?
3b) can you create surfaces on multiple independent graphics cards? [ie non SLI]
performance wise it might not work well, but one of my ideas is to create a shim that opens multiple directX surfaces on 2 or more graphics cards (or even the same one) but tells the game it's got one surface, then it routes the directX calls to the appropriate card that's actually doing the rendering... there's probably be all sorts of timing and synchronization issues, but it might be possible (if I understood directX better).