PyOpenGL on Linux: waiting for hardware acceleration

[srenner at lycosmail.com]

Does anyone have hardware acceleration working with Linux yet? I have
been waiting for over a year -- it is always reputed to be just
around the corner. If anyone wants to do complete 3d toy world/game
prototyping in Python, the tools are already in place. I should know,
since that's what I've been spending all my time on.

Anyone who has tried this sort of thing knows that framerate is
absolutely crucial. When it drops too low, it is just not possible
to navigate and see what is going on. Mesa's software renderer leaves
most of my little demos fill rate bound. When Python is the slow part,
judicious dumping of geometry into display lists speeds things up again.

Supposedly Xfree86 4.0 is due out soon and will have hardware support.
Please email me if you know how this will work. In particular, will
display lists (in Mesa) be passed to the hardware renderer? Or will
the board just do fast primitives, leaving all transforms to Mesa? (This
point isn't directly relevant to PyOpenGL, but I'd like to know.)

If you've read this far, you might be interested in doing an open source
collaborative project. Here are some ideas:

*Various projects based on my attention-renderer, already released in
the PyOpenGL demo directory. (There's a new release coming soon). This
already implements some new ideas, like infinitely detailed objects.
(You know, fractals.) 

*How about a ray-tracer or scanline renderer in pure Python? It would
be dead slow but interesting to write, especially the shaders. I've
rejected this before because my realtime renderer seemed more
interesting, but if others want to work on this I say let's do it. (Did
I mention radiosity?)

*If you happen to have a retinal display with eye movement tracking JUST
LYING AROUND please mail me IMMEDIATELY.

sr


--== Sent via Deja.com http://www.deja.com/ ==--
---Share what you know. Learn what you don't.---