DirectX usually has the ability to emulate features that aren't available through hardware. This comes at the cost of using the CPU to emulate the features, which typically is going to come with a heavy performance hit.
I'm not sure you quite understood what I wrote. I've always been able to get COMPLETE DirectX11 acceleration with everything (tessellation, compute shaders, everything), no mather whether RDP or console. I can always query the device and context (without a swap-chain), no matter what scenario with DirectX11, it always works correctly. Any emulation is out of question (not only because I see I'm getting the same device with the same drivers and parameters and everything but also the performance with enormous loads is exactly the same).
So I was hoping somebody would tell my why this isn't the case with OpenGL. I
do know that vendor drivers just hook into the OpenGL1.x dll from Microsoft and provide OpenGL (1.5, 2.x, 3.x, 4.x, extensions, whatever) functionality this way. Is there any reason why this shouldn't work without a console session?
I'm no expert on Windows drivers and administration so I don't see why the hook doesn't "kick-in" via RDP or telnet or whatver, too. I'm only assuming it's the creepy OpenGL context creation mechanism (which, for example, prevents us from initialising OpenGL easily in no-window scenarios).
I wish I'm wrong.
