Thank you very much for your help.
Julia
--- On Wed, 8/19/09, Eugene Loh <eugene....@sun.com> wrote:
From: Eugene Loh <eugene....@sun.com>
Subject: Re: [OMPI users] MPI loop problem
To: "Open MPI Users" <us...@open-mpi.org>
List-Post: users@lists.open-mpi.org
Date: Wednesday, August 19, 2009, 12:29 PM
Julia He wrote:
does the problem show up even if you turn MPI off?
My answer: Yes and No. Yes, when I used ifort compiler. But the problem
did not show up when I used gfortran, which was much slower.
Okay. So, if this is not an MPI problem, as much as we like you and
wish you luck, this mail list might not be the best place to resolve
this problem! :^)
Humor aside, there appears to be something special about how the code
is compiled, and the code looks like it's non-trivial. So, you
probably need to pursue this issue somewhere other than this mail
list. I don't know what resources are available to you for pursuing
ifort issues.
Since you have a working and non-working version, maybe there are ways
of "divide and conquer" to isolate what the essential difference is
between the two. E.g., does the ifort version work at lower
optimization levels? Etc.
Good luck. I'm sorry, but this mail alias probably isn't the right
place to resolve this particular problem.
So, I think I can conclude that the problem is related to
the Intel compiler -- ifort. I guess the mpif90 and mpif77 are wrappers
to ifort.
(I should know this, but OMPI man pages should probably tell you for
sure.) I suspect the mpi wrappers can be used to invoke "any"
compiler. In any case, it appears you've isolated the problem to be
due to the underlying compilers and not to an MPI component.
Do you know how to solve this problem? Is there any flag
that I need to use in ifort?
Sorry, I'm not familiar with ifort.
-----Inline Attachment Follows-----
_______________________________________________
users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users
