[Cryptography] floating point

[Tom Mitchell]

     There are multiple choices...
Some would do well to code so they can test some options.

https://software.intel.com/en-us/blogs/2008/03/06/intel-decimal-floating-point-math-library

Consider that 64 bit integer arithmetic has a nice dynamic range and is
easy to extend to
128 or even 256 bits at the expense of RAM.   Compact data structures like
IEEE floating point are best addressed
in hardware but other floating point representations are available and
worthy of testing
for some problems.    My favorite was the Motorola floating point library
for the MC68000
that was put forward before the IEEE standard became soup.

Also transcendental functions like logarithms and trigonometric values are
notorious for
being "sloppy" again the quality of these values is a rich topic of
research... because the
compromises for reason of speed etc. that are taken in lib math
(libm.so+math.h) are not
always as good in some cases as the old CRC printed tables.


-- 
  T o m    M i t c h e l l
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20141228/e383d297/attachment.html>