Re: Reduction operations in P1788/D9.5
On 2015-01-11 21:45:44 -0500, Michel Hack wrote:
> Lundi 12 Janvier 2015 à 03:43:41 +0100, Vincent Lefèvre a écrit:
> > On 2015-01-11 09:07:43 -0500, Michel Hack wrote:
> > > Note that there are few genuine error cases in 754 -- the only one
> > > that comes to mind (besides SNaN conversion) is sqrt() of a negative
> > > argument.
> >
> > What do you mean?
>
> You left out the explanation:
>
> >> The "indefinite forms" 0/0 or Inf-Inf can in many contexts
> >> be interpreted as B, because the zero or Inf may have been "impure",
> >> i.e. the result of underflow or overflow.
I now understand what you mean (I thought you meant that there was
a wrong choice in IEEE 754). But I disagree. First 0/0 can also be
regarded as an error case (if the zeros are actually exact). There
are other functions that can generate a NaN, such as log, rSqrt,
powr, asin, acos, acosh, and atanh.
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)