Re: RENDER ext. & hardware acceleration, libart integration

Peter Kaczowka <[email protected]> writes:

> Without understanding Render too well, I would guess that some Render
> operations could be sped up using MMX.  I'm not volunteering to do it, but for
> anyone interested I think you'll find that using SIMD and coding MMX is great
> fun.  But then again I'm old school, and coding assembler still strikes me as
> more fun than say Java.
> 
> If MMX is already being used in Render or has already tried, I'm sorry to imply
> that no one has thought of this before.

I have some code for MMX acceleration of common Render operations
around, that I keep meaning to clean up and give to Keith. :-(

(And there is a lot of optimization of the software fallbacks that can
be done without reference to MMX... you can get maybe 5x improvement
over the current code with plain C, and then MMX gives you an extra 50%.)

The hardware I've tested on doesn't seem to do burst-mode reads from
video memory, so there is limit to how well you can do when
software compositing directly to the frame buffers. There are definitely
big  improvements by carefully managing when things are in video memory, 
when in software, and when in both places.

Regards,
                                        Owen