---
0. IMPORTANT... why are you doubling the "even" terms individually?  STUPID!
   - make it so you have four new macros that use an additional 3 carry variables
        - SQRADDSC - store first mult      [ simple store, no carry ]
        - SQRADDAC - add subsequent mults  [ 3n word add ]
        - SQRADDDB - double the carry      [ 3n word add ]
        - SQRADDFC - forward the doubles into the main [ 3n word add, note, x86_32 may need "g" instead of "r" ]
   - only use the four macro pattern for rows with >= 3 "doubles"
        - otherwise use the existing SQRADD


1. Write more documentation ;-)
2. Ports to PPC and MIPS
3. Fix any lingering bugs, add additional requested functionality.
4. Unrolled copies of montgomery will speed it up a bit
5. 


NOTE:  The library is still fairly new.  I've tested it quite a bit but that doesn't mean surprises
can't happen.  Please test the results you get for correctness.