DConf 2014 Day 2 Talk 3: Designing an Aurora: A Glimpse at the Graphical Future of D by Adam Wilson

via Digitalmars-d-announce digitalmars-d-announce at puremagic.com
Tue Jul 8 22:30:19 PDT 2014


On Wednesday, 9 July 2014 at 04:26:55 UTC, Tofu Ninja wrote:
> Modern graphics hardware has a very well defined interface and
> all modern graphics api's are all converging on matching the
> hardware as close as possible. Modern graphics is done by 
> sending
> buffers of data to the card and having programmable shaders to
> operate on the data, period.

That's true, but OpenGL is being left behind now that there is a 
push to match the low level of how GPU drivers work. Apple's 
Metal is oriented towards the tiled PowerVR and scenegraphs, 
probably also with some expectations of supporting the upcoming 
raytracing accelerators. AMD is in talks with Intel (rumour) with 
the intent of cooperating on Mantle. Direct-X is going lower 
level… So, there is really no stability in the API at the lower 
level.

But yes, OpenGL is not particularly suitable for rendering a 
scene graph without an optimizing engine to reduce context 
switches.

> largely artificial. In modern graphics api's the difference
> between 2D and 3D is merely a matrix multiply. If the immediate
> mode was designed how I suggest above, then 2D vs 3D is a non
> issue.

Actually, modern 2D APIs like Apple's Quartz are backend 
"independent" and render to PDF. Native PDF support is important 
if you want to have an advantage in the web space and in the 
application space in general.

There is almost no chance anyone wanting to do 3D would use 
something like Aurora… If you can handle 3D math you also can do 
OpenGL, Mantle, Metal?

But then again, the official status for Aurora is kind of unclear.


More information about the Digitalmars-d-announce mailing list