FWIW: I just cloned the v3.0.x branch to get the latest 3.0.1 release
candidate, built and ran it on Mac OSX High Sierra. Everything built and ran
fine for both C and Fortran codes.
You might want to test the same - could be this was already fixed.
> On Dec 11, 2017, at 12:43 PM, Ricardo
Hi guys
I’m having problems with a Fortran based code that I develop with
OpenMPI 3.0.0 on Mac OS X. The problem shows itself with both gfortran
and intel ifort compilers, and it runs perfectly with version 2.1.2 (and
earlier versions).
Launching the code, even without using mpiexec, causes
As you note, there is no correlation between release number and libtool
versioning. The comment in VERSION explains the libtool rules and you can see
why the two values differ.
The PMIx master is at release version 3.0 as it includes new APIs that have not
yet been released. The configure
I made a good deal of progress, and I now have OpenMPI 3.0.0 capable
of being run.
I do have one final point of confusion, however. It appears that
--with-pmi= fails because configure is testing for
a non-existent major version of PMIx, 3. Perhaps that is the major
version of the OMPI
Your code looks correct. There are few things I would change to improve it:
- There are too many calls to the clock(). I would move the operations on
"time" (the variable) outside the outer loop.
- Replace the 2 non scalable constructs to gather the 2 times on the root
either by an MPI_Reduce(+)