Fractal Speedup Question
Paul Zimmermann
Paul.Zimmermann at loria.fr
Tue Jan 17 23:39:31 CET 2006
> Question: can this speed up the calculations and how can I use fixpoint
> numbers with GMP ?
Usually for fixed-point calculations it is faster to use the mpz layer.
A fixed point number f is represented by an integer m with an implicit
exponent e:
f = m / 2^e.
A floating-point addition is converted to an integer addition, a floating-point
multiplication to an integer multiplication followed by a division by 2^e
(mpz_div_2exp).
Paul Zimmermann
More information about the gmp-discuss
mailing list