By my reading this maps conventional fixed-function OpenGL onto a
shader
implementation.
Yes, exactly.
As you implied, we don't have sufficient hardware to
implement a fully programmable shader, so it isn't a lot of use.
We do not have enough to implement all the features present in pre-
G80 cards, yes.
This does not stop us from making a fast, limited capability
processor like those in the
G80, and continue to parallelize them untill we run out of silicon.
That is the major advantage
of this architecture - even if we only have 4 shaders, it will still
run (albeit very slowly).
I wouldn't be surprised if this is how fully programmable sharer
hardware (eg.
the G80) works. If a fixed-function OpenGL path is used, the driver
just
uploads a suitable shader function to the hardware.
After reading about the G80 a little, that is almost exactly what I
was thinking of.
Once we synthesize our DMA/VGA cpu core, we can get a reasonable
estimate of the % silicon
needed for a simple pipelined RISC and go from there, unless somebody
already knows.
Hope that clarifies my thoughts!
Nicholas
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
