Re: [boost] [SoC] multi precision arithmetic library proposal.

From: Guillaume Melquiond (guillaume.melquiond_at_[hidden])
Date: 2006-05-01 10:40:39

Le lundi 01 mai 2006 à 13:42 +0200, Stefan Slapeta a écrit :

> IBM and IEEE furthermore try standardize some (special) decimal floating 
> point types for future hardware support (see n1977!), but from what I've 
> heard, this seems to be rather chaotic and controversial in IEEE.

The situation is not as bad as it may seem. Formats, arithmetic
operators, rounding modes are already defined, are not controversial,
and will be part of the revision of IEEE-754 standard. So a library can
be written now without risking from becoming unusable once the standard
is published.

The only controversial point at the moment is the binary low-level
encoding of the numbers (a hardware-efficient encoding would be really
inefficient for pure software implementations). So from a library point
of view, the only difference between the various proposals lies in the
accessor functions: reading/writing mantissas/exponents of decimal
floating-point numbers.

Best regards,

Guillaume