[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: g77 problems
- From: Richard Henderson <rth cygnus com>
- To: axp-list redhat com
- Subject: Re: g77 problems
- Date: Thu, 3 Sep 1998 17:41:39 -0700
On Thu, Sep 03, 1998 at 06:23:55AM +0000, Wes Bauske wrote:
> > What happens with numbers which are so small/large that they can not
> > be normalised (sorry I can't remember the term); as far as I understood
> > they were valid numbers, but perhaps I'm confused.
>
> Underflows are set to zero according to the architecture
It's called a "denormal".
Underflows _may_ be set to zero. In that there is an fpcr bit that
one may set to enable this. If it isn't, the processor traps for
software emulation of the denormal, which of course looses if you
havn't compiled with -mieee.
> If a number cannot be represented normalized, you're in
> trouble because IEEE FP format has an implied 1 bit in the
> mantissa.
No, exponent == 0 and mantessa != 0 is a denormal, and does not
have the implied 1 bit.
r~
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
[]