Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

From: Richard Wait (Richard.Wait_at_[hidden])
Date: 2006-05-10 03:41:02


Quoting "Jeff Squyres (jsquyres)" <jsquyres_at_[hidden]>:

> If you are looking for the path of least resistance, then going back to
> MPICH is probably your best bet (there is certainly merit in "it ain't
> broke, so don't fix it").
>

True - but where is your sense of adventure!!

> However, there may be a few other factors to consider:
>
> - Just because an app runs and completes with one MPI implementation
> does not mean that the application is correct; running with a different
> implementation can be a great way to find bugs that you were unaware of
>

This is precisely the situation I found since at three places in the code I had
       ...
       call MPI_TYPE_COMMIT(side,ierr)
       ...
       call MPI_TYPE_FREE(side)
       ...

 Neither MPICH nor OpenMPI object on compilation but, as I mentioned MPICH
 runs but OpenMPI needs the correct syntax
       ...
       call MPI_TYPE_FREE(side,ierr)
       ...
 to run correctly.

Thanks for email.
 Richard Wait
 Uppsala University
 Department of Information Technology
 Scientific Computing
 Box 337
 75105 UPPSALA
 Tel: 018-4712757
 Fax: 018-523049

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.