Really, AMD HD5000 or Nvidia GT300 series hardware is the minimum necessary to perform hardware tesselation with OpenGL?
AFAIW AMD introduced the hardware tesselation OpenGL extension with the HD3800.
Is the Phoronix article right? Someone knows for sure?
There is tesselation hardware in the older cards, but it isn't the same as what the newer cards have (and is less powerful). The demo requires the new tesselation hardware, because it's based on the DX11 requirements which the old unit can't support.
Think of it like going from SM2 to SM3. The older cards had shaders, but a new demo might still require the newer version.
I wonder whether the OpenGL extension that gives access to tesselation on the HD5xxx series is the same one as the currently experimental extension that is used for tesselation on the HD3xxx and HD4xxx series cards. I know the hardware is different, but that doesn't necessarily means that the same OpenGL extension can't be used.
The OpenGL requirements for tesselation could end up differently than the DirectX requirements for tesselation.
Anyway: good to hear we'll have it with Catalyst 10.1. I'm looking to buy myself a new and shiny HD5870 around that time .
I have a question, because the marketing materials say all kinds of stuff, how is it really with new features in graphic cards?
It seems that in PC games openGL is non-present, and new GPU features are always connected with directx. AFAIK, all things that work for DX work for opengl one way or other.
How are GPU manufacturers designing these things?
So they implement it in HW and then it can be reached with drivers?
I read somewhere that something like this tesselation can be achieved with much older HW, but it wouldn't be that effective.
That is supposed to be a problem with rasterization, that little can be changed in a GPU pipeline from developers perspective.
Demonstration of capability is the reason it is important. Unigine has for their entire existence ensured that the Unigine is a cross platform engine.
This actually opens up their market beyond games into visualization and media production - have a look at their partner page on , gaming isn't the only market that needs good visuals.
Sounds like an good way for linux to get more momentum in the graphics department. the demo is impressive the onsite portofolio did not match that imho.
This ofcourse didn't answer the second part of my question. Searching the site and some other sources explained it well enought. Crossplatform support still takes good planning and considerable ammount of extra work on the graphics part (besides all the other platform dependend stuff).
Don't see why you should be locked to the newest cards unless you want to you the latest features of the engine... thats has nothing todo with crossplatform or not.
this engine is as crossplatform as the come.
A "platform" consists of the operating system, the software installed and the hardware installed. Hence if you only port over some different OS but not different hardware then you are not really cross platform. The term is often though used in a very loose way to hype a product which is why nowadays suddenly tons of engines call themselves cross platform but actually are not ( or only partially, which though is better than nothing ).