| Mike Zulauf 2004-12-23, 4:07 pm |
| NOTE, I already posted this message on some other groups. I meant to
cross-post it here as well, but it slipped my mind. Sorry for any
inconvenience. . .
---------------------------------------------------------------------
Hi all,
I have a 3D large-eddy simulation code designed for atmospheric
modeling. It uses MPI and domain decomposition to run on parallel
clusters.
Presently, the dynamics are compressible, but I'm investigating the use
of a Poisson type solve to obtain pressure (anelastic dynamics). In the
distant past I've coded this sort of thing up myself, but that was for
serial code. I'm not really in the mood to try and code it up myself
now, and I'm sure others have already done it (and better than I could).
What sort of solutions have people seen for this? It looks to me, at
first glance, that PETSc may be able to handle it. Anybody have
experience with this?
Any other recommendations?
My code uses horizontal domain decomposition (x and y) - but not
vertical (z). Are any given libraries designed to work with that sort
of decomposition "out of the box", or would I need to do any sort of
remapping?
The code is in Fortran, so the routines would need to be
callable/linkable with that.
Any recommendations, comments, etc?
Thanks,
Mike
--
Mike Zulauf
mazulauf@met.utah.edu
|