MITgcm configs: Difference between revisions

From Fluids Wiki
Jump to navigation Jump to search
(Rewrite to stop referring to config files that are stored in this wiki. The latest ones are in the UW git repo. Explain a bit better what they're for and how they get used.)
 
(3 intermediate revisions by one other user not shown)
Line 1: Line 1:
== MITgcm Configuration Files ==
== MITgcm Configuration Files ==


* for bow, minnewanka, waterton
The MITgcm source code needs to be compiled with settings appropriate for whichever platform it will be run on.
** this configuration file expects the SGI MPT module and GNU compilers
Each platform has a corresponding configuration file.  These configuration files are included in the MITgcm bundle you pull
** be sure to specify '''-mpi''' on your '''genmake2''' command line
from the UW git repository. You'll see them under <code>/path/to/MITgcm/tools/build_options/</code>. The file names reflect
** [[bow_mpt|bow_mpt]]
the hostnames of the servers.


* for hood.math
The <code>makempi.sh</code> script uses these configuration files and sets appropriate flags and load modules in order to compile MITgcm on the Fluids Group's servers and on Compute Canada (a.k.a. The Alliance) servers.
** this configuration file uses the Intel compilers and OpenMPI 1.10
** [[hood_mpi_intel|hood_mpi_intel]]
 
For other machines, see Prof. Lamb.

Latest revision as of 15:08, 13 May 2024

MITgcm Configuration Files

The MITgcm source code needs to be compiled with settings appropriate for whichever platform it will be run on. Each platform has a corresponding configuration file. These configuration files are included in the MITgcm bundle you pull from the UW git repository. You'll see them under /path/to/MITgcm/tools/build_options/. The file names reflect the hostnames of the servers.

The makempi.sh script uses these configuration files and sets appropriate flags and load modules in order to compile MITgcm on the Fluids Group's servers and on Compute Canada (a.k.a. The Alliance) servers.