How bad is fast-math?

[Jean Francois Martinez]

On Wednesday 30 May 2001 01:59, Michael Meissner wrote:
> On Wed, May 30, 2001 at 01:35:27AM +0200, Jean Francois Martinez wrote:
> > On Wednesday 30 May 2001 01:02, Michael Meissner wrote:
> > > On Wed, May 30, 2001 at 12:25:50AM +0200, Jean Francois Martinez wrote:
> > > > According to the doc fast-math generates incorrect code.   Has the
> > > > situation changed?   In case it has doc should be updated, in case it
> > > > hasn't name should be changed and become something derogatory since
> > > > there are people who see the fast in fast-math and use it without
> > > > further investigation.   It is being used in aLinux distribution and
> > > > I have seen it recommended in a Linux magazine
> > >
> > > In general, it depends on what you consider the correct code to be :-)
> > >
> > > If your code never encounters infinities and denormals, and never
> > > passes an out of range argument to an intrinsic function, then
> > > -ffast-math will certainly generate faster code on some machines.  I
> > > can't say how well that matches up with the numeric code out there.  I
> > > would imagine many codes don't stray into these dark corners (but then
> > > again, many do go into that territory).
> >
> > Use in glibc would be certainly foolish.   And I think it is being done.
>
> If it is being done in an unsafe manner, then you need to bring it up with
> the glibc developers in the proper forum.

It is notr the glibc people, it is a well known linux vendor who uses 
fast-math in its default compil parms like evidenced by looking in the 
/usr/lib/rpm directory.  I haven't checked if it does the same thing for 
glibc but I think so.

			JFM