New acceleration architecture

Adam Jackson ajax at nwnk.net
Sun Jun 26 19:15:21 PDT 2005


On Sunday 26 June 2005 13:44, Alan Cox wrote:
> That seems silly. There are several cases where angled lines are
> horrendous worst cases for screen fetch/screen put operations especially
> as Xorg generally doesn't have DMA screen tile fetch so your performance
> for a fetch is a joke. Vertical lines in software also cause tlb
> thrashing. So yes that one does matter as any xtank player would
> understand 8)

If you're drawing the line to the backing store and blitting that to the 
screen, then there's no fetch.

> > 6) The code as presented in the snapshot will be checked in on Monday
> > morning/Sunday night, dependently on the feedback I'm going to get.
>
> I'm watching with interest. The Voodoo2 has hardware assist for software
> compositing in 2D so this may help make better use of it. (Right now I
> don't boot the 3D engine as thats horrendously complex to use).

While we're on the subject, the voodoo driver's xaa Render accel doesn't get 
used in 6.8 or later due to the new blend hooks, and when I ported it  - 
basically verbatim - to the tdfx driver it failed rendercheck.  Looked like 
simple premultiplied-alpha confusion but I never followed up on fixing it.

The way you did it in voodoo(4) is... unique, but entirely reasonable given 
how atrocious xaa is.  The major win for kaa (and exa, it looks like) is that 
you can cache Render pictures offscreen as well, which means subsequent 
blends don't require an upload, just a command push.  The ability of the 3dfx 
hardware to do alpha blends through the 2d pipe means we should be able to 
get Render accel without having to expand them into POT textures.  It's just 
enable alpha blend, set factors, and blit.

- ajax
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.x.org/archives/xorg/attachments/20050626/442597b7/attachment.pgp>


More information about the xorg mailing list