Valve To Support DX10 With Episode 2 96
In an interview with Game Informer from last week, representatives from Valve confirmed that they'll be supporting DirectX 10 functionality in the release of Half-Life 2: Episode 2 and Team Fortress 2. This will be the case even for those folks who haven't upgraded to Vista yet. No worries if you don't have a DX10 card, though. They've got functionality nailed all the way back to DirectX 8, and are trying to push it all the way back to 7.
Re:DX10 on Windows XP? (Score:3, Informative)
Re:opengl? (Score:1, Informative)
Re:Great but... (Score:5, Informative)
Unreal 3 is openGL hence why more companies are using that compared to Valve's Source engine. Hopefully they will get the hint sooner rather then later.
You do realize that PS3 and the like use OpenGL ES, which is NOT the same as the GL on computers?
Besides, they have tons of custom extensions necessary to use these machines efficiently...
Oh, and forget about a Source rewrite for OpenGL. There just is not point in this. Direct3D works on the platform 96% of all PC gamers use. A rewrite is EXTREMELY time-consuming, because of the differences in the API designs. We're talking about at least a 6-month-delay here (very likely more).
Both DirectX and openGL just tell the gfx card what to do.
But not equally. GL binds sampler states to a texture, D3D binds them to sampler stages. D3D has +Z as "inward", OpenGL -Z. D3D has +Y as "down", OpenGL "up". There is no equivalent to an OpenGL rectangle texture in Direct3D. The GLSL API works quite differently than the HLSL one etc. Do you want to finance the rewrite, the bug-fixing, beta-testing?
The fact that they decided to use DirectX which only works on Microsoft platforms for a game engine they're trying to license to other companies is pretty stupid from a business point of view.
"Only" is quite funny. Windows is an enormous gaming platform. Also, you get Xbox support nearly for free. As for machines like the PS3 and the Wii, forget about having one universal engine for all of them. ALL AAA titles are written specifically for one title, and maybe ported to another, requiring substantial rewrites (this is why usually console titles arent ported to other consoles). Try porting Shadow Of The Colossus from PS2 to Wii for example.
Sorry, but your suggestions are absolutely suicidal for all but the wealthiest of all game development companies. Because of the ARB being much too slow, OpenGL stagnated in the important years 1996-2001. Heck, a decent render-to-texture mechanism got introduced 2005, while DirectX already had one 1998. OpenGL was in an excellent position back in the 90s: Direct3D 3 sucked, OpenGL was better, easier, finer. But if you have graphics card manufacturers and game developers on one side, demanding more features, and an obscenely slow ARB on the other side, there can be only one solution - create another API. Nowadays many codebases are D3D based precisely because OpenGL just sucked in the post-D3D7 era. And rewriting the entire codebase is suicide, as already said. Which is a shame, because OpenGL is pretty decent now, and if the OpenGL 3 rumors are right, it will rock.
The biggest deals (Score:3, Informative)
2) Unified shader API. All shaders (pixel, vertex and geometry) are talked to in the same fashion in DX10. Makes for easier design. However it also allows unification on the back end. Though it isn't required, as a practical matter the shaders will be unified on the cards. The GeForce 8800, the only DX10 card out, has unified hardware shaders and ATi's R600 will as well when it hits the market. This means that shaders can be tasked to different things as needed. If a scene is complex pixel shading wise but simple vertex wise, no problem, the shaders can do that, and then switch around the very next scene.
3) Geometry shaders. DX9 didn't support them, and DX9 class hardware doesn't have them.
4) Support for video memory virtualization and preemptive multitasking of the GPU. Basically giving the GPU to really share its resources effectively.
Re:Great but... (Score:3, Informative)
First, I'm assuming you meant to say that all AAA titles are written specifically for one platform, etc. Assuming that is what you meant, I also think this is a pretty faulty statement. Have you looked at console gaming lately? The majority of titles out there appear on at least two consoles, if not all three. The latest iteration of the Call of Duty franchise would be a good example. It can be found on the Xbox and Xbox 360 and PS2 and PS3 and Wii. Clearly it's not that onerous to port from one system to another, if the makers of A and AA titles are willing to spend the money to do so. They're clearly getting back more money than it costs them to do the port, probably many times over, otherwise they wouldn't keep doing it.
The reason AAA titles don't get ported has very little to do with technical details, and a whole lot to do with marketing and business strategy. In the console world, developers making a AAA title shop their project to the various console manufacturers, and hope to make it an exclusive for one of those systems. In return, the console make cuts them a nice fat check. That check buys them the time they need to spend on polishing the game in order to make it a AAA title. In the case of Valve, there is no real competition in the PC gaming space, so there's no manufacturers to play off each other in order to get a check for making your game exclusive. But Valve doesn't really need the money anyway, they've got quite a nice little warchest as it is. They could easily spend the money to either hire talent or outsource production to allow ports of their games. However, they don't want to give up creative control, or have some semiautonomous internal division produce a half-assed port. Basically, they don't want to risk "diluting their brand" through ports. Either way you slice it, though, it's a marketing/business thing, not a technology thing.