> The beauty of the idea? One card renders
> the odd scan lines, the other the even. Some rocket scientist
> should win a Nobel for that one.
Until a black-bag organization finds out how to reintroducedithered patterns.
> What this all means is that you can trade off refresh
> and frames for quality of rendering and detail.
Or buy RAM and work in 3D. Speaking of which quality:
> Now, I'm not recommending shorting Intel...it's just...
> why would all these CPU producers
> start putting graphics instructions on the CPU?
Because programmers, sFx wizards or not, can't handle, track
and document more than 16 bits of precision through the said
80% of code used to eat the OS's/screen-API's lunch.
It's possible that the spot artists are hosing the CASE systems
used to build their APIs, or that compiler (really the project
manager package's bits in BUILD) has bugs that are more
affecting than usual, but 'wider is better' fails ubiquity.
How is this? Let's have a look at the candidates for Boyce &
Diprima's CD! They were many, some packaged with CORBA
1.0 gadgets, some with pari (aka gp, neoMathCad) mutations,
with Wolfram Software losing money testing them. The hand-
coded ones had the best of the (losing, all) lot. All the tools
are better, but 'pass value through entity' is still a tough row to
hoe. Perhaps the IETF-draft to code and symbol-to-code
sectors aren't close enough to the corpus callosum?