Re: [Gnash-dev] cairo instead of agg for fb?

[Timothy Lee]

 Dear all,

I was the one who initially wrote the Cairo rendering backend.  It 
currently relies on Cairo to perform tessellation.  Speed could be 
improved by feeding polygons directly to Cairo.

Regards,
Timothy Lee

On 09/08/2009 11:28 PM, Leon Woestenberg wrote:
> Hello Udo,
>
> thanks for replying and your work on AGG and FB in Gnash!
>
> On Tue, Sep 8, 2009 at 9:32 AM, Udo Giacomozzi<address@hidden>  wrote:
> > Monday, September 7, 2009, 10:36:49 PM, you wrote:
> > LW>  what is needed to support cairo (with it's image surface) for the
> > LW>  framebuffer UI?
> >
> > LW>  Currently only agg is supported. Cairo / pixman recently got good ARM
> > LW>  NEON optimizations that are worth exploring.
> >
> > I wrote the AGG and FB part of Gnash, but don't know much about Cairo.
> > However, as long as Cairo can provide a linear buffer that can be
> > copied directly to graphics memory, it should not be a big deal. There
> Yes, that's no problem with cairo.
>
> > are various pixel formats and -depths around, so Cairo must support
> > all of these directly - otherwise you will loose a lot of performance
> > just for converting pixels.
> >
> > The ARM NEON feature sounds very interesting. AGG (the library itself,
> > not the Gnash GUI) could certainly be optimized to use these features,
> > so maybe you might invest some time there...
> I would rather spend time on enabling cairo in the fb mode, so that we
> get the NEON optimizations for free.
>
> Note however that Cairo may be a tad slower to begin with, so it has
> to overtake the agg performance.
>
> Regards,