Messages in this thread | | | Date | Sun, 07 Feb 1999 19:11:17 -0500 | From | Ben Bridgwater <> | Subject | Re: Linux Graphics Architecture - later, I guess |
| |
Geert Uytterhoeven wrote:
> On Sat, 6 Feb 1999, Ben Bridgwater wrote: > > The only other base of preexisting (or soon to be) code is the XFree 4.0 > > acceleration engines, which would be fine too, although they might take more > > effort to repackage, and that project appears to be understaffed as is. > > If they're undermanned, why don't you join them?
Apart from lack of time, I think it'll be a *long* time until there's a portable XFree-based solution to what I want to do (direct graphics writing from an X client). Even if/when they provide direct rendering for local X clients, they're still going to have to tackle the fundamental issue of multiple programs (in this case multiple X clients plus the server) contending for the graphics hardware. If this contention were solved with mutexes in the rendering engine, then that would probably require multi-threading in the server to overcome the client fairness problems is would create...
I'm rather puzzled why there appears to be so little interest in what seems to be the obvious solution of Linux adopting graphics device drivers, and solving the contention issue once and for all. I can understand purists who say "all Linux graphics should be via X", but a device driver approach would help X purists just as much (client rendering, server porting, platform stability, etc.) as people who have other needs for it.
Presumably with the release of 2.2 and introduction of framebuffers, this hardware contention is going to become much more of an issue. I'm guessing there will be people who will actually want to *use* the framebuffer in parallel with X, not just have it "mode switch compatible"... and there's only going to be one way to fix those problems! ;-)
Anyways, I guess I'll stick to DGA for the time being, until the contention/portability issue becomes more urgent.
Ben
Please CC: bennyb@ntplx.net
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |