This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: Infamous PR fortran/19925
- From: dominiq at lps dot ens dot fr (Dominique Dhumieres)
- To: fortran at gcc dot gnu dot org
- Date: Mon, 15 Sep 2008 14:26:57 +0200
- Subject: Re: Infamous PR fortran/19925
> ... saving off a pointer to the end of this list should cure the CPU time.
I am not sure that will be enough. If I read the code correctly, there are
two passes: one to build the list (for which your proposal will change the
quadratic behavior to a linear one) and one to "evaluate" it starting from
the end. For the later I think a pointer to the ancestor will also be required
("double-linked list?) to get a linear behavior.
Now constructors are intended to build vectors, using a list to descibe them
does not seem to be the right solution (it may explain the large prefactor
I have found for pr34554).
Dominique
PS I think this thread should be redirected toward pr19925.