graphics hardware; OpenVG

[Shawn Rutledge]

On 2/4/07, Jon Phillips <jon at rejon.org> wrote:
> Well, you would want to add openvg support lower than X11. You should
> talk to the X and Cairo developers about this.

Well I wondered how a standard like that can improve application
portability.  QT, X, and OpenGL (and GTK) are all somewhat portable
themselves, and applications are usually written on top of those
layers.  The first three can run directly on hardware (or maybe GTK
can too nowadays?)  But when you choose one of those you limit the
applications that can run on the device to those that were written
against that API.  By introducing a new standard, it means that if
applications are to use that new standard directly, they have to be
re-written; and if not, then the new standard is just another layer of
indirection between one of those existing systems and the hardware.
At least it defines which kinds of acceleration are the most important
and helps to quantify the results.  I do think that having accelerated
drawing hardware for more types of primitives than just line segments
would be a good improvement (elliptical arcs and Bezier curves are
important for many kinds of graphics; and there is a second-order
curve that is important for TrueType, right?)  But there were existing
attempts like GGI to make accelerated drawing available at a lower
level.

>
> For the OpenMoko uses, Cairo offers hardware acceleration through Glitz,
> soooo....and Cairo has been in GTK since 2.8...

But how soon do you think we will see portable Linux devices that
provide acceleration for most of OpenGL?  I have a suspicion that
Apple already had to do that, to be able to get that glitzy UI to work
well enough, just like desktop OS X will not run very well without a
good graphics card.  And there are some chips available.  I think I
read something about an Intel-designed companion graphics chip for the
PXA270.  And Nvidia has something too.  I wonder what the power budget
is for those.