This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [Fortran] Another array reference dependency improvement
- From: Steve Kargl <sgk at troutmask dot apl dot washington dot edu>
- To: Paul Thomas <paulthomas2 at wanadoo dot fr>
- Cc: Roger Sayle <roger at eyesopen dot com>, fortran at gcc dot gnu dot org, gcc-patches at gcc dot gnu dot org
- Date: Sat, 4 Mar 2006 09:40:44 -0800
- Subject: Re: [Fortran] Another array reference dependency improvement
- References: <Pine.LNX.4.44.0603040807230.14380-100000@www.eyesopen.com> <4409CF1F.5050302@wanadoo.fr>
On Sat, Mar 04, 2006 at 06:32:15PM +0100, Paul Thomas wrote:
> Roger,
>
> >This latter approach avoids the need to pass down a linked list
> >of variables through gfc_dependency's call stack, and then scan
> >this list everytime we encounter a variable at a leaf.
> >
> >Perhaps the gfortran maintainers could indicate which approach
> >they think is most appropriate.
> >
> >
> I favour the latter - ie. a context flag. In doing the Polyhedron tests
> to check out your first patch, I notice that compilation times are 50%
> up on 6 months ago; there's too much running up and down linked lists!
>
I agree with Paul. I also vaguely recall that the index of an
implied-do-loop needs some special handling, and the latter approach
may be applicable there. Unfortnately, a quick scan of standard has
jogged my mind about the details of implied-do-loops.
--
Steve