At 14:15 23/07/98 +0100, you wrote:
>
>>Yes but if you don't use VDI calls you've got to make a version for each
>>diferent video card you could found on atari computer. May be these
>>optimisation could be donne directly in VDI (ie in Fenix for example).
>
>
>If you DO use VDI calls, then every complex primitive (like gouraud, phong
>or textured polys) will have to be drawn using a VDI setpixel function. That
>will be so slow, it will defeat the point of having the capability in the
>first place.
Yes it's right :-(
>
>No amount of optimisation will make the setpixel call fast enough, because
>the polygon rendering needs an integrated span-rendering ability that does
>not pass through an individual setpixel. In other words, if you wanted to
>support it quickly through the VDI, you would have to add the full opengl
>drawing routines to the VDI itself. We are not really able to do that.
Yes but it could be integrated in Fenix for example or in all the new VDI
that are in dev (even in Milan Tos 4.05 in the future).
>
>I think it would be good to support VDI as the base compatibility device
>context - but a secondary accelerated context would be required for most
>practical applications. Supporting the most common colour depths would
>not pose too much of a problem.
Yes, using fast optimised routines in a virtual frame buffer the use vdi to
blit it in screen windows.
>
>Almost all colour formats fall into these categories:
>
>8-bit mono
>8-bit RGB standardised (rainbow) truecolour palette
>8-bit indexed palette
>15-bit 1555 RGB
>16-bit 565 RGB
>24-bit 888 RGB
>32-bit 8888 ARGB
>
>Anything else is normally a byte-swapped (intel) version of the latter 4
>formats. I think supporting all of these would be possible without too much
>trouble.
>
>There are examples in the MESA/OpenGL source that use mixed software
>and hardware (VDI/GDI style) calls to get the best of both worlds. But most
>of the acceleration is gained from line/rectangle GDI functions. Polygons
>will only be accelerated if there is 3D hardware available.
>
>Our TOS 3D acceleration layer could be based on optimised 3D routines
>for the Falcon/Atari platform, and possibly even true 3D hardware support
>for S3 Virge cards etc. There is a Virge driver with MESA, so that is a
>start.
>
Yes so we need PCI bus on AB040... and i haven't finish the pinout trace ;-))
Rodolphe.
NO'MAD SARL / STraTOS
62, Bd Aristide Briand
77000 Melun FRANCE
SIRET : 411 413 826 00016
RCS Melun B 411 413 826
FAX : (+33) 164 378 228
HTTP://www.chez.com/stratos/
Received on to. juli 23 1998 - 21:57:00 CEST