Aaron:

This new version of tiegcm is called tgcm14a. It runs on
Crays and SGI O2K. The mpi stuff is NOT ready -- it is
being developed for the IBM SP's -- but will be ignored
by the Makefile when running under IRIX.

I have put 2 tar files on anon ftp at ftp.hao.ucar.edu
in /pub/foster:

-rw-r--r--   1 foster   ftp      73809920 Mar  3 09:23 tgcm14a_data.tar
-rw-r--r--   1 foster   ftp      1239040 Mar  3 09:28 tgcm14a_src.tar

The data file contains a netcdf primary history file, which
you can use as a SOURCE file for input. It has 5 histories on
it (1,0,0 to 5,0,0). The data tar file also has 2 netcdf files 
for gpi and magnetic field data. All netcdf files have a
".nc" suffix.

The source tar file was made as follows:

tar cf tgcm14a_src.tar Makefile Depends Objects mkdep \
  tgcm.setup tgcm.exec *.h *.f *.F *.inp *.ncd

The Makefile is a gnu file (I use gnumake on our o2k).
The Makefile includes the files Depends and Objects, which list
depedencies for the make. These 2 files can be generated
with the perl script mkdep, also in the source tar file.
Any time you change the code in a way that will affect 
dependencies, just run "perl mkdep" to refigure the 
dependencies before remaking -- it handles both include and 
use statements.

I included the tgcm.setup and tgcm.exec scripts only for
your information -- these scripts are used when we submit
production runs -- you shouldnt need to execute them, but
check them out, esp the env vars that are set for IRIX
in tgcm.exec.

There is a sample input file (tgcm14a.inp I think), and
the *.ncd file is an ncdump of the history file, as folllows:

ncdump -v"lon,lat,lev,mtime,year,day,ut,timestep,f107d,f107a, \
  hpower,ctpoten,byimf,mag,dtide,sdtide,colfac,p0,nflds"

As before, I have not tested this on non-ncar machines --
obviously you may have different compilers, etc, and I may
have forgotten to include something. Let me know how it goes...

--Ben

Let me know when you get the tar files, so I can remove the
big one from the hao disk.