Angus Dorbie (dorbie++at++sgi.com)
Wed, 08 Sep 1999 21:57:05 -0700
It's just GFX hardware. Lot's of bandwidth to GFX helps.
>
> >> So, what about images resident in RAM but bigger than texture memory?
> >
> >You can always add texture memory management to your app and subload to
> >the graphics over time.
>
> But I don't want to do that! That incresases cost, and more importantly
> _RISK_.
> I'm preparing a tender, so I'm interested in how I can make my offer more
> attractive basically `for free'.
OK the simplest way is to oversubscribe and just page. Anything more
will require some coding but will give better performance. The
oversubscription will likely be highly dependent on driver quality
issues and have unacceptable performance glitches.
This is all referring to the use of many images which total > texture
memory, not a single image which is > texture memory.
>
> >> [BTW, is anybody working on an OpenGL extension for texture compression,
> >> now that there are chips supporting the Direct3D stuff?]
> >
> >It's actually pretty straightforward to support this using new internal
> >format tokens. The only question is do you compress on download or
>
> I just found this:
>
> http://www.s3.com/s3tc/s3tcoglext.htm
Looks like that's exactly what they have done.
This spec does seem to have a severe flaw. The pixels format says that
pixels should contain the untiled compressed data. This is the crux of
the problem and it seems to mix both worlds in a highly confusing way.
If the data is compressed you need to define the color endpoints for
each cell somewhere then the smaller 2 bits per pixel as the cell
pixels. The natural way to do this is to pass table entries then packed
bits for each tile. If you don't then you need to spec where the cell
color values are specified and where the pixel table entries are,
unfortunately this isn't explained but it is essential to the operation.
Shouldn't be too difficult to clarify things though.
Cheers,Angus.
--
"One of the best-known folk theorems of software engineering is that
60% to 75% of conventional software projects are either never
completed or rejected by their intended users. If that range is
anywhere near true (and I've never met a manager of any experience
who disputes it) then more projects than not are being aimed at goals
which are either (a) not realistically attainable, or (b) just plain
wrong."
Eric S. Raymond - The Cathedral and The Bazaar
For advanced 3D graphics Performer + OpenGL based examples and tutors:
http://www.dorbie.com/
This archive was generated by hypermail 2.0b2 on Wed Sep 08 1999 - 21:57:07 PDT