[cairo] rookie question

Surith cairo at surith.net
Fri Oct 10 23:06:53 PDT 2008

> It does seem strange to me that there always seems to be excitement
> about using OpenGL to improve our 2D rendering stack

Some of the excitement you are referring to is likely coming from
embedded developers.

Open GL ES is the only viable accelerated graphics API for many
embedded devices today. It's somewhat widely available : Nokia's
Series 60, various Linux based devices, the iPhone and even some
Windows Mobile devices. Unfortunately, driver support on all but the
iPhone has been rough around the edges.

The alternatives I know of truly suck : DirectX Mobile implementations
are worse than useless, small X servers are good for demos but haven't
really been accepted by the users and developers. I don't know enough
about DirectFB, which seems to have been occasionally discussed in the
Cairo roadmaps.

There seems to be potential here. Simple hardware acceleration results
in big performance boosts on embedded platforms : in previous tests
with a proprietary 2D renderer, accelerating only the blend routines
resulted in about 15% faster performance in mildly complex, animated
test cases.

(BTW, are lower performance improvements expected on desktop
platforms? Is this why there seems to be skepticism from some? Or are
Win32/Quartz/X fast enough as it is? Or is implementing an Open GLES
backend expected to be too much work?)

Cairo is already a good solution for embedded devices. (Thanks to
everyone's work!)
If driver support improves and if an Open GL ES backend for Cairo
demonstrated improved performance from the current image backend, it's
going to get a lot of uptake from embedded developers.

 - Surith

More information about the cairo mailing list