fortran parallel-netcdf
John Tannahill
tannahill1 at llnl.gov
Mon Jul 28 19:22:14 CDT 2003
Robert,
I don't understand the comment below about having to include pnetcdf.inc;
I am pretty sure that it was already included in the test code I sent you?
I have trimmed my test codes down even further until all they are supposed
to do is create a netcdf file and define a single dimension and variable.
The C version works fine; the Fortran version creates the file, but the
file itself is empty. I have included both codes and their Makefiles again.
Still running things on NERSC's IBM SP2 (seaborg).
Also before I worked on the above, I grabbed and installed the 0.8.6
version. Install on seaborg was easier:
gunzip parallel-netcdf-0.8.6.tar.gz
tar xf parallel-netcdf-0.8.6.tar
cd parallel-netcdf-0.8.6
module load gnu
setenv CC xlc
setenv FC xlf
setenv MPICC mpcc_r
setenv MPIF77 mpxlf_r
autoconf
./configure --prefix=/usr/common/homes/j/jrt/Pnc/parallel-netcdf-0.8.6
vi macros.make
Change:
CC = mpcc_r
CXX = xlC_r
FC = xlf
To:
CC = mpcc_r
CXX = mpCC_r
FC = mpxlf_r
gmake
gmake install
I made two versions with the only difference being that I used the
--with-fortran=yes flag for the library I built for linking with my
Fortran test code? Not sure whether or not I need to set all four
environment variables in the beginning? Also, looks like I still
should be editing macros.make?
Anyway thanks for all your help. I am at your beck and call as I
cannot make much progress until I am able to utilize the Fortran
interface. Please keep me updated. Perhaps Richard Hedges or Tyce
Mclarty here at LLNL can take a crack at it as well.
Regards,
John
Robert Latham wrote:
> On Mon, Jul 28, 2003 at 01:25:15PM -0700, John Tannahill wrote:
>
>>Robs,
>>
>>Have you been able to reproduce my problem? Can I be of any help on this
>>end? I am in kind of a holding pattern until this is resolved.
>
>
> I've gotten it to build once i included pnetcdf.inc (otherwise,
> nf_clobber would be undefined).
>
> When i try with the 'testfs' and 'pvfs' filesytem drivers, everything
> seems to run ok for me on Chiba city with the portland group
> compilers.
>
> I'm not sure what to say, other than i need a bit more information
> about how your fortran program fails.
>
> ==rob
>
--
============================
John R. Tannahill
Lawrence Livermore Nat. Lab.
P.O. Box 808, M/S L-103
Livermore, CA 94551
925-423-3514
Fax: 925-423-4908
============================
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: csnap.c
URL: <http://lists.mcs.anl.gov/pipermail/parallel-netcdf/attachments/20030728/061ebf8b/attachment.diff>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Makefile.c
URL: <http://lists.mcs.anl.gov/pipermail/parallel-netcdf/attachments/20030728/061ebf8b/attachment-0001.diff>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: pnf_test.F
URL: <http://lists.mcs.anl.gov/pipermail/parallel-netcdf/attachments/20030728/061ebf8b/attachment-0002.diff>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Makefile.F
URL: <http://lists.mcs.anl.gov/pipermail/parallel-netcdf/attachments/20030728/061ebf8b/attachment-0003.diff>
More information about the parallel-netcdf
mailing list