ffpa math libs?

[Scott Moody]

We are interested to know how the math libraries work for the sun.
Specifically, how does using the different floating point options on 'cc'
affect the -lm math library?  My understanding is that to use the 'fpa'
and '68881' chips, the 'c' compiler must generate in-line code for those
chips. How does the math library do this, since that same math library
also works on machines without these chips. 

Is there a switch statement in the math library that calls the 68881 math
routines?  If so (and it looks like there is), is there any way to get the
math library to call the 'fpa' math routines instead? Does the fpa include
math routines? If not they what good is the fpa when we do lots of math
calls, other that the normal *,/,+,-  ??

Thanks for any responses?

 Scott Moody @ Boeing Mountain Network
     ..uw-beaver!ssc-vax!shuksan!scott