Re: New Math Coprocessor - who has a need for it???

[kentm@pro-applejacks.cts.com (Michael Kent)]

dempson@swell.actrix.gen.nz (David Empson) writes:

>QuickDraw II: No.  ACE: No.  Other compression algorithms: possibly,
>but I doubt it (I think some really advanced and really slow
>algorithms use floating point, but they aren't common).
>
>Pointless: possibly.  It depends on whether its font scaling uses
>fixed point or floating point arithmetic.  My guess is fixed point, in
>which case the NC won't make any difference.

Since Pointless and many common compression programs (Shrinkit, AutoArk,
and HardPressed) are all currently supported, would the authors of these
programs be willing to support the card?  I'm sure Henrik would be willing
to modify GSymbolix to support it (am I right?), and Lim Thye Chean (I hope
I didn't mutilate that too badly) would probably modify his Mandelbrot
programs.  I'm assuming that supporting the Number Cruncher wouldn't be too
hard to do, but many of the recent number-crunching wares are still
supported.  We could make this work.

Of course, one question is: Would a compression routine using FP on a Num-
ber Cruncher be faster than using INT on the 65816?  Any thoughts, Andy? 
Either of you?

Mike