[MITgcm-support] MITgcm viz

[Ed Hill]

On Tue, 2004-08-17 at 16:33, Michael Schaferkotter wrote:
> greetings:
> i want to setup MITgcm on a platform that doesn/t have matlab.
> generally our models have platform independent binary i/o capabilities,
> so we can move the binary data back to a system and process it.


Hi Michael,

The MITgcm model is designed to work nicely on either big-endian or
little-endian architectures in part because the MDSIO-based read and
write routines all have the ability to do byte-swapping (or not!) during
the binary file reads and writes.  Thus, binary files produced on one
platform can be read (without any changes) on another platform no matter
what the native byte orders on the two systems happen to be.

Similarly, the newer MNC I/O interface also supports transparent
byte-swapping since it is automatically handled by the NetCDF library.


> is there an example that shows the use of the mnc package?

The verification example "exp0" was recently modified to test MNC so it
contains a complete MNC setup.  But please be aware that not all of the
model outputs have been converted to use MNC/NetCDF.  We (mostly me) are
still busy converting all of the possible model outputs to be run-time
switchable between the newer MNC and the older MDSIO mechanisms.


> is mnc supported under MPI?

Yes, MNC will run nicely on MPI systems and this has been tested and
verified on our Intel Linux cluster, on two Altix systems, and on a DEC
Alpha.  There are no conflicts between MNC and MPI (or pthreads for that
matter) because all I/O is done in a per-tile fashion.

If you have any problems with MNC, please send bug reports to this list
and I'll do everything I can to see that it works for you!

Ed

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464