MITgcm

• runs on a wide range of computer systems.
• includes special software (called WRAPPER) to support efficient execution and relatively easy portablity.

• Unix PC's ( both Intel and AMD processor based) running Linux.
• Unix workstations from Sun, SGI, HP, IBM, Compaq.
• Unix based PC and workstation clusters.
• high-end symetric multi-processor systems.

Under the WRAPPER, code can also run very efficiently on high-end vector processing systems, but we can't afford one.

• a computer runing a Unix style operating system.
• a Fortran 77, Fortran 90 or Fortran 95 compiler.
• a C compiler.

• a multi-process computer

and either of
• a compiler and that supports multi-threaded programs.
• a system with an implementation of the MPI library installed.


Instructions on running multi-process codes can be found here. Instructions regarding muti-threaded execution can be found here. Specialized communication software, to optimize performance on a specific platform, can also be introduced into the WRAPPER. This process is discussed here. Before running a parallel code we recommend you try one of the basic examples

• an automatic differentiation tool.

An example experiment illustrating the automatic differentiation process can be found here.

Once you have your computer set up

go here to find out how to download the model. The full distribution (including examples with input datasets and sample output) is around 50MB. Instructions on estimating the computing resource requirements (CPU time, memory usage and disk space) for a set of numerical experiments can be found here.

Simulation output is usually examined

using one of several standard analysis tools, depending on individual preference and experiemnce.

Still confused?

send us mail. We like our software to be used and are happy to help out if we can.

---

Last modified on $Date: 2001/10/21 20:50:25 $

CVS: /u/gcmpack/mitgcm.org/sealion/download_software/viaCVS.html,v $Revision: 1.6 $