Re: FPE Demand?

[o9938156@unicorn.it.wsu.edu (mark lyon)]

Nathan Mates (nathan@cco.caltech.edu) wrote:
: In article <HOCKLEYD.95Jan20112215@dec3300.cuug.ab.ca>,
: Dallas Hockley <hockleyd@dec3300.cuug.ab.ca> wrote:
: >I too would shell out between $100-$200 based on how fast and how compatible
: >the FPE would be.  The ORCA compatibility would almost be a necessity.  Unles
: >you can give an equivalent solution in effectiveness and ease of use.

:    Orca compatability? Orca won't be sped up much at all by a
: _Floating_Point_ accelerator. (Read: it'll take the sine of 2.5 faster
: than unaccelerated.)

It depends on how many floating point operations are done while
compiling a program ;)

:    If you mean compatability with Orca compiled stuff, it should be
: transparent with everything that does things the legitimate way.
: (FTA demos and such won't be accelerated since they don't use floating
: point).

It would be transparent - provided the SANE toolset was patched in
some way - with previously compiled stuff.  I think what was meant
is this:  there is a compiler pragma that tells Orca to generate
the code to call the FPE directly, bypassing the tools entirely,
and it would be nice if the new card were compatible with this
pragma, so an update to Orca would be unnecessary.  

: Nathan Mates

Mark Marr-Lyon
o9938156@unicorn.it.wsu.edu