Generic get_d_2exp failures

paul zimmermann Paul.Zimmermann at inria.fr
Sat Oct 28 20:16:14 UTC 2017


       Niels,

> It would be nice if we could find a portable way to add two floating
> point values without rounding up.

see http://perso.univ-perp.fr/langlois/slides/imacs05_sl.pdf,
slide 10, algorithms "Two Sum" and "Fast Two Sum" in the literature.

Paul


More information about the gmp-devel mailing list