whoops! need to decrement by a bigger value indeed!! 
it was all fine and dandy til i tried compiling the math library with 
it.. 
there was a 10E20 constant somewhere in the src, and the compiler had to 
slog through it, and after i heard the disk stop grinding for an eerily long
time, i went over to see it happily churning through about 10 billion 
iterations of that loop.. not good! :)

here is what should be a much better version, it gets right down to what 
needs to be decremented. it could be made even shorter, i will do that once
i sit down and think about it.. this below is a gross hack but it is better
than the even grosser one before!

isildur


/* new version */
ENTRY(modf, 0)
        cvtdl   4(ap),r2                # get the integer part out
        jvs     1f                      # integer overflow
        cvtld   r2,*12(ap)              # put it in float format, in dest
        subd3   *12(ap),4(ap),r0        # put fraction in r0
        ret
1:
        movd    4(ap),-8(fp)            #copy original value
        extzv   $7,$8,4(ap),-20(fp)     #get exponent and sign bit from orig
        ashl    $-7,-20(fp),-20(fp)     #kick it down for a subtraction
        decl    -20(fp)                 #subtraction
        ashl    $7,-20(fp),-20(fp)      #kick it back up to its place
        movd    $0f1.0,-16(fp)          #1.0, floating
        bisl2   -24(fp),-16(fp)         #fast floating multiply
        subd2   -16(fp),-8(fp)          #decrement it to be close to 1
        cvtdl   -8(fp),-24(fp)          #convert to integer
        cvtld   -24(fp),-16(fp)         #get the remaining integer part
        subd3   -16(fp),-8(fp),r0       #put fraction in r0
        subd3   r0,4(ap),*12(ap)        #put integer component back
        ret