long doubles

Torbjorn Granlund tege at swox.com
Wed Feb 8 15:54:45 CET 2006

Daniel J Farrell wrote:

  Maybe because long double is hardware specific? i.e. some time is 12  
  bytes long other times its 16. Also, different manufactures divide  
  the number of bytes up differently between exponent and the number  
  part of the floating point.
Right on the money.

Damien Stehle wrote:
  Besides, why does mpz_set_d rounds towards zero and not to the nearest
  (with the round to even rule)?

Because that's analogous to how double -> int conversions are
done in C.


More information about the gmp-discuss mailing list