[Cfp-interest 1673] Re: intmax_t removal
Paul Zimmermann
Paul.Zimmermann at inria.fr
Fri Jun 26 06:04:12 PDT 2020
Dear Mike,
> (Some say that integer
> types are needed for efficiency, and that was true 40 years ago, but on
> 32-bit+ architectures today integer multiply and divide are often slower
> than the floating-point operations...)
then I am part of "some". To implement arbitrary precision arithmetic,
20 years ago some people did use floating-point types because you could
have 53 bits per word instead of 32 bits, and the efficiency was comparable.
Nowadays nobody uses floating-point types anymore to compete with GNU MP.
And even for floating-point computations, using integer types might be
faster [1,2].
Best regards,
Paul
[1] https://hal.archives-ouvertes.fr/hal-01227877
[2] https://members.loria.fr/PZimmermann/talks/quad.pdf
More information about the Cfp-interest
mailing list