Re: RENDER ext. & hardware acceleration, libart integration

Keith Packard <[email protected]> Sat, 24 Aug 2002 15:29:15 -0700
Newsgroups gmane.comp.xfree86.render
Message-ID <E17ijP9-0000oZ-00@localhost>
Around 17 o'clock on Aug 24, Owen Taylor wrote:

> While I'm no PCI bus expert, my conclusion when I fooled around
> with this area some is that the video card manufacturers simply
> don't spend any money/effort on the read-from-video-memory part of 
> their PCI hardware. 

They don't have to bother -- the CPU makes all frame buffer reads 
synchronous.  I believe DMA to be significantly faster; it might be worth 
DMA'ing the whole image, doing the composite, and DMA'ing it back.

Keith Packard        XFree86 Core Team        HP Cambridge Research Lab