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

[kpopple@imp.sim.es.com (Ken Poppleton)]

In article <CA8u44.5II@wave.scar.utoronto.ca>, 90taobri@wave.scar.utoronto.ca (TAO BRIAN T) writes:
> In article <1993Jul15.090209.10130@ifi.unizh.ch>, gudat henrik writes...
> > 
> > QuickDraw and I guess ACe as well use integer math. Only floating point
> > stuff will be accelerated.
> 
>     Why can't math copros and DSP's speed up integer math as well? 
> Sure, FP math is slower, but int math is used in more places.
> --
> Brian Tao:: taob@io.org (Internex Online, 416-363-3783, 10 lines, v.32bis)
> ::::::::::: 90taobri@wave.scar.utoronto.ca (University of Toronto, 9T4)
> 

Yep, the DSP will improve integer math and also fixed point math.  But
it will need software to do the math along with the patches for the SANE
library.  This is in contrast to the 68881 based coprocessors which only
need the SANE patches.  

THe DSP board is progressing slowly, development of the DMA interface
is taking longer than I had planned primarly because I only have 1 to 2
hours per week to work on this project.  Is there anyone working on or
thinking of software for the DSP board?  If I am the only one developing
software for the DSP board, then it will be a long time before a math
support library is completed (That is not on the top of my list of
software to write).

-----------------------------------------------------------------------
Ken Poppleton 
kpopple@imp.sim.es.com
-----------------------------------------------------------------------