Re: DRI integration
Keith Packard <[email protected]>
| Newsgroups | gmane.comp.freedesktop.xserver |
|---|---|
| Message-ID | <[email protected]> |
Around 22 o'clock on Nov 13, Keith Whitwell wrote: > It would certainly be cleaner to have everything go through GL, though. Certainly the goal should be for the bulk of rendering to be accelerated. > As the nature of 2d rendering changes and potentially looks more like > traditional 3d rendering - lots of geometry, texture or other high-bandwidth > needs, forinstance, the old results may no longer apply. We'll see where 2D applications go. Right now, they're still drawing very little on the screen and using a lot of static images and text. It seems clear though that this may well change in the future. What I'd like to do is to build an accelerated cairo implementation on GL to provide a comfortable PS/PDF rendering model for 2D application developers. That may make all of our discussions about how to best accelerate X operations moot; 'X' applications may end up just using GL for all of the performance sensitive drawing where available and fall back to software on embedded systems that don't have GL support. Cairo is much closer to GL than Render; although we would like to continue to provide per-component compositing for improved text appearance on LCD screens; and perhaps even extend that to geometric objects as well. -keith