On Apr 3, 2006, at 1:31 PM, Bryan Follis wrote:
> checking for g77... no
<snip>
> checking for lf95... no
> checking for gfortran... gfortran
> checking whether we are using the GNU Fortran 77 compiler... yes
> checking whether gfortran accepts -g... yes
> checking for ranlib... ranlib
> checking for mpi.h... yes
> checking for MPI_Initialized in -lmpi... yes
> checking if have Fortran MPI bindings... no
> checking if supposed to have Fortran MPI bindings... yes
> configure: error: Cannot continue
>
> Now, it is a problem that gfortran is being seen as the GNU Fortran
> 77 compiler, right? It is supposed to be an F90 or F95 equivalent
> compiler, correct? Could this be the cause of the errors where it
> dies at the MPI bindings? Are there any build options I can do when I
> initially build LAM-MPI or for the lamtests that will work around
> this problem?
The problem is not that gfortran is being used as the Fortran 77
compiler - this is perfectly fine, and we do that all the time with
LAM/MPI.
The output from configure is kind of weird - it's doing a compiler
search, meaning it didn't find our mpif77 installed. We override
configure's default compiler search order to check for mpif77 first
(similar to mpicc and mpic++). Can you see if there is an mpif77
executable in the bin/ directory of your LAM/MPI installation? If
not, somehow you have an incomplete installation of LAM/MPI. If so,
make sure it is in your path and executable (ie, the command "mpif77 -
showme" should work).
Brian
--
Brian Barrett
LAM/MPI developer and all around nice guy
Have a LAM/MPI day: http://www.lam-mpi.org/
|
