OpenGL vs. DirectX on embedded GPUs

Hi All,

As you know OpenGL on Intel embedded GPUs has some problems that forces many vendors to switch off hardware acceleration on these machines (running Windows).

Does it happen also with DirectX or with DirectX you can take the advantage of hardware accelertion even on these embedded GPUs?

And more generally why there is this lack of compatibility between OpenGL and the embedded GPUs?

Thanks,

Alberto

Embedded GPUs are usually low end manufactured by non-3D graphics cards vendors. Therefore their GL drivers can be unreliable or lagging behind the current version. GL drivers are actually implementation, unlike Direct3D, which requires alot of effort to make, which can be an obstacle for these vendors where they are not specialized in the 3D area.

Thanks glfreak,

Should we expect that all the embedded GPU behave in the same way or some of them are better then others?

Thanks,

Alberto

Embedded GPUs from Nvidia seem less problematic than others on the GL front.

AFAIK Intel GPUs are the most common. Usually embedded chips are less capable than installable cards, and hence even if they have reliable drivers, they will not be able to support higher versions that demand features which are not available in those embedded hardware.

As I mentioned before, GL is an implementation which require considerable effort and dedication to come up with a reliable and stable implementation vs. Direct3D approach of separating drivers from implementation.

I wish we could promote OpenGL to this approach and have the ARB/Khronos become responsible for the core implementation, and let IHVs make drivers and extensions.

I bet this is the solution rather than pushing IHVs to create reliable “drivers,” and bug them all the time by not well supporting GL.

I think Mac OS adopted this approach and therefore it has the GL as the reliable standard for it’s 2D/3D graphics.

So Direct3D is less problematic, isn’t it?

What are pros and cons of the two different approaches?

Thanks,

Alberto

On the driver side yeah.

D3D approach makes it easy for IHVs, unlike GL’s where IHVs have to “implement” the API rather than writing minimal set of core hardware functionality exposed by their drivers to D3D, where it uses it’s API on top of this functionality.

Now I see,

So Direct3D is not affected by embedded GPUs limitations?

Thanks so much for your help.

Alberto

It’s still affected, but there’s always a reference/software layer at least, and it’s very unlikely you will get pink/black textures if you requested a feature that’s withing GL 1.2 spec. In short drivers work better. Wanna support more hardware and lower end, go for D3D

Now everything is clear to me.

Thanks so much again.

Alberto

I found an interesting dicussion here on the subject and want to know your opinion:

http://www.solidsmack.com/solidworks-opengl-direct3d-gpu-comparison-cad/2008-09-01/

Thanks,

Alberto

The thoughts there are mostly based on the dark ages of GL1.4-2.0, imho. Would a CAD user use a laptop with integrated Intel GMA or Radeon? Need to use Quadro to accelerate CAD apps? Then why is Maya running as fine on a GF (http://www.techarp.com/showarticle.aspx?artno=539&pgno=8)?