two questions on GMP & C float
tg at gmplib.org
Sat May 16 18:16:36 CEST 2009
Marc Glisse <marc.glisse at normalesup.org> writes:
On Sat, 16 May 2009, Sylvain Pion wrote:
> interval arithmetics library is switching FPU rounding mode
> down. Does GMP use FPU or may I leave rounding mode set to "down"
> for the interval arithmetics library to work even during calls to
I believe that in the few cases where GMP uses floats, it only uses
exact computations on those and the rounding mode is irrelevant (this
may not be true if you have some non-IEEE double type not using base
2, but I don't expect that to be the case).
This is mostly true, but not entirely. I think we have a few cases in
the assembly code, e.g. computations of inverses using Newton's method,
where we get rounding.
More information about the gmp-discuss