Parallelization of GISS Model E: MPI etc.
                Igor Aleinov

I would like to talk about general issues of parallelization of GISS
Model E.  I will start from an overview of the hardware and of typical
methods  for writing parallel code (i.e. MPI and OpenMP). I will then
describe  the implementation of MPI inside Model E, show some scaling
results  and discuss possible ways for future development. The goal of
this talk is to get everybody up-to-date with respect to current state
of Model E and to help developers to write their code in
MPI-compatible  way.