A New Acceleration Architecture For X

[phoronix]

Phoronix: A New Acceleration Architecture For X

XAA, or the XFree86 Acceleration Architecture, is over twelve years old and finally in 2005 it was greeted by a replacement, EXA. XAA is nearing an end-of-life and Intel is prepared to remove XAA acceleration within their next Intel graphics driver release later this year...

http://www.phoronix.com/vr.php?view=NjY0MA

[FunkyRider]

He's better make it really, really good. Otherwise he looks like a jerk to me, always wiping out other's code using his code

[Regenwald]

mh, i'm getting some troubles. what about glucose? is exa/glucose/uxa usable with gallium3d? or are they going to be obsolete and will be substituted? why do they have such big plans when it doesn't makt *that* much sense to put lots of efforts into the "old" dri-model?

[oyvind]

Xorg development looks very chaotic from the outside. Can't be much fun to program GFX device drivers using a framework in complete limbo all the time. I'll go with nothing but NVidia's proprietary solution, since they pretty much rip out DRI, and it works so blazingly well (and they are working on the 2D-problems for GF8000/9000-gen cards). When will the free next-gen-DRI stack ever mature ? Let's hope the over-night trashing-and-introducing-a-new-API process works out well in the end for *all* graphics-card brands (not just Intel). However long that will be.

[puelocesar]

Well, if that means that 2d acceleration in Linux will not suck anymore, I'm in all for it!

[Pickup]

Let me see: Intel had a memory manager done by Tungsten graphics (TTM) and then they threw it away and replaced with their own GEM. After 12 years we finally have a better 2d acceleration architecture, and just the time for it to be ready and Intel threws in a replacement. And now anyone who was going on working to EXA and TTM based drivers must stop and redo all the work from start.
Which technology will be shot dead next? Gallium3d looks the best candidate in the line.
Going on this way, X.org may
1) become an Intel-only graphix server or
2) fork (again) into an Intel-X and exa-ttm-gallium-Y.

[remm]

Actually, it does more look like TTM took forever and got nowhere. GEM is already virtually included in the kernel. Similarly, EXA was not that successful, it seems ...

Gallium3d could indeed be next in line, because it also failed to deliver for the time being Is it really something bad, I don't know.

Obviously, Intel is running the whole show to their benefit, as they're the only one of the big 3 that is directly contributing development resources.

[Melcar]

Well, if it's "better" then why not. Still, all this wait is really annoying.

[madman2k]

Let me see: Intel had a memory manager done by Tungsten graphics (TTM) and then they threw it away and replaced with their own GEM. After 12 years we finally have a better 2d acceleration architecture, and just the time for it to be ready and Intel threws in a replacement.

Its not like intel throws everything existing away. Both GEM and UXA are close to TTM and EXA. While the former can be seen as a subset of TTM the latter is just an enhanced version.
Better they do it now, than in 2 yers when all drivers have already been ported.

[Pickup]

Actually, it does more look like TTM took forever and got nowhere. GEM is already virtually included in the kernel. Similarly, EXA was not that successful, it seems ...

Gallium3d could indeed be next in line, because it also failed to deliver for the time being Is it really something bad, I don't know.

Obviously, Intel is running the whole show to their benefit, as they're the only one of the big 3 that is directly contributing development resources.

But then, piece by piece, the whole X will be rewritten around Intel's hardware needings.
Will this drive the other manufacturers to Intel's solutions or will they say "Ah well, let Intel get those leenookz losers, we won't support anything but windoze anymore".
Will Intel be the only available choice for linux and (maybe also) Opensolaris users?
Is there a chance Intel will seize the X code and change the license?