Charm++ parallel programming system

[Sam White <white67 AT illinois.edu>]

Yes, only 1-3D chare arrays are currently supported in f90charm. I see no philosophical reason for these restrictions (no singleton chares, groups, or nodegroups), it is simply an implementation detail. With modern Fortran and some development effort, f90charm could definitely be extended and improved. But, as far as I know, f90charm has not been touched in several years as our group's efforts at targeting Fortran have been invested in AMPI.

-Sam

On Mon, Jul 11, 2016 at 9:04 AM, Jozsef Bakosi <jbakosi AT gmail.com> wrote:

Hi Sam,

I'm also interested in more information regarding Fortran support.

In particular, the Fortran90 bindings manual says "you cannot define Chare and Group types". Does that mean, Charm++ groups are not supported using Fortran?

Thanks,

Jozsef

On Sat, Jul 9, 2016 at 9:39 AM, Sam White <white67 AT illinois.edu> wrote:

Hi Niel,

Charm++ provides limited support for Fortran90, and you can find the manual below. The example f90charm programs in charm/examples/charm++/f90charm/ all compile and run, but this interface has, I think, seen very limited use. Adaptive MPI, our implementation of MPI on top of Charm++, provides Fortran bindings too.

http://charm.cs.illinois.edu/manuals/html/f90charm/manual.html

-Sam

On Fri, Jul 8, 2016 at 3:58 PM, Carlson, Neil <nnc AT lanl.gov> wrote:

Hello,

Your FAQ page indicates limited support for Fortran and suggests
contacting this email for further information.

I was turned on to Charm++ through an enthusiastic colleague who has
been experimenting with it in some research C++ codes.  My interest is
in large-scale multiphysics (PDE) codes (e.g., https://github.com/truchas)
I'm interested in undertaking a serious investigation of charm for those
types of algorithms, however I am unabashedly a Fortran person and
would like to discuss with someone where charm stands with respect
to Fortran and whether it would really be a viable path to pursue.  My
interest is in Fortran 2003/2008 and its object oriented features.

Thanks,
  Neil

---
Neil Carlson, Ph.D.
Telluride Project Leader
Computational Physics and Methods (CCS-2)
Los Alamos National Laboratory
PO Box 1663, MS D413 · Los Alamos, NM 87545
505.665.6386