| 1 |
$Header$ |
$Header$ |
| 2 |
|
|
| 3 |
|
|
| 4 |
MITgcmUV Getting Started |
MITgcmUV Getting Started |
| 5 |
======================== |
======================== |
| 6 |
|
|
| 7 |
o Introduction |
o Introduction |
| 8 |
|
|
| 9 |
o References |
This note is a guide to using the MIT General Circulation Model Ultra Verstaile |
| 10 |
|
implementation, MITgmcUV. MITgcmUV is a Fortran code that implements the |
| 11 |
|
algorithm described in Marshall et. al. 1997, Hill, Adcroft, ... |
| 12 |
|
The MITgcmUV implementation is designed to work efficiently on all classes |
| 13 |
|
of computer platforms. It can be used in both a single processor mode |
| 14 |
|
and a parallel processor mode. Parallel processing can be either multi-threaded |
| 15 |
|
shared memory such as that found on CRAY T90 machines or it can be multi-process |
| 16 |
|
distributed memory. A set of "execution enviroment" support routines are |
| 17 |
|
used to allow the same numerical code to run on top of a single-process, multi-threaded |
| 18 |
|
or distributed multi-process configuration. |
| 19 |
|
|
| 20 |
|
o Installing |
| 21 |
|
To setup the model on a particular computer the code tree must be created |
| 22 |
|
and appropriate compile and run scripts set up. For some platforms |
| 23 |
|
the necessary scripts are included in the release - in this case follow |
| 24 |
|
the steps below: |
| 25 |
|
|
| 26 |
|
1. Extract MITgcmUV from the downloadable archive |
| 27 |
|
tar -xvf checkpoint12.tar |
| 28 |
|
|
| 29 |
|
2. Create platform specific make file |
| 30 |
|
For example on a Digital UNIX machine the script "genmake.dec" can |
| 31 |
|
be used as shown below |
| 32 |
|
|
| 33 |
|
cd bin |
| 34 |
|
../tools/genmake |
| 35 |
|
cp Makefile.alpha Makefile ( On Alpha machine) |
| 36 |
|
|
| 37 |
|
3. Now create header file dependency entries |
| 38 |
|
make depend |
| 39 |
|
|
| 40 |
|
4. Compile code |
| 41 |
|
make |
| 42 |
|
|
| 43 |
|
5. Copy input files |
| 44 |
|
cp ../verification/exp2/[a-z]* ../verification/exp2/*bin . |
| 45 |
|
|
| 46 |
|
6. Run baseline test case |
| 47 |
|
setenv PARALLEL 1 |
| 48 |
|
dmpirun -np 2 ../exe/mitgcmuv ( Under Digital UNIX ) |
| 49 |
|
mpirun.p4shmem ../exe/mitgcmuv -np 2 ( Under Solaris + mpich) |
| 50 |
|
|
| 51 |
|
|
| 52 |
|
This runs a 4 degree global ocean climatological simulation. |
| 53 |
|
By default this code is set to use two processors splitting |
| 54 |
|
the model domain along the equator. Textual output is written |
| 55 |
|
to files STDOUT.* and STDERR.* with one file for each process. |
| 56 |
|
Model fileds are written to files suffixed .data and .meta |
| 57 |
|
These files are written on a per process basis. The .meta |
| 58 |
|
file indicates the location and shape of the subdomain in |
| 59 |
|
each .data file. |
| 60 |
|
|
| 61 |
|
o Running |
| 62 |
|
|
| 63 |
|
- Input and output files |
| 64 |
|
|
| 65 |
|
Required files |
| 66 |
|
============== |
| 67 |
|
The model is configured to look for two files with fixed names. |
| 68 |
|
These files are called |
| 69 |
|
"eedata" and "data". |
| 70 |
|
The file eedata contains "execution environment" data. At present |
| 71 |
|
this consists of a specification of the number of threads to |
| 72 |
|
use in X and Y under multithreaded execution. |
| 73 |
|
|
| 74 |
|
- Serial execution |
| 75 |
|
|
| 76 |
|
- Parallel execution. Threads |
| 77 |
|
nSx, nSy |
| 78 |
|
setenv PARALLEL n |
| 79 |
|
nTx=2, nTy=2 |
| 80 |
|
|
| 81 |
|
- Parallel execution. MPI |
| 82 |
|
mPx, nPy |
| 83 |
|
dmpirun |
| 84 |
|
|
| 85 |
|
- Parallel execution. Hybrid |
| 86 |
|
|
| 87 |
|
o Cutomising the code |
| 88 |
|
|
| 89 |
|
Model structure |
| 90 |
|
=============== |
| 91 |
|
The "numerical" model is contained within a execution |
| 92 |
|
environment support wrapper. This wrapper is designed |
| 93 |
|
to provide a general framework for grid-point models. |
| 94 |
|
MITgcmUV is a specific numerical model that uses the |
| 95 |
|
framework. |
| 96 |
|
Under this structure the model is split into |
| 97 |
|
execution environment support code and conventional |
| 98 |
|
numerical model code. The execution environment |
| 99 |
|
support code is held under the "eesupp" directory. |
| 100 |
|
The grid point model code is held under the |
| 101 |
|
"model" directory. |
| 102 |
|
Code execution actually starts in the eesupp |
| 103 |
|
routines and not in the model routines. For this |
| 104 |
|
reason the top level main.F is in the eesupp/src |
| 105 |
|
directory. End-users should not need to worry about |
| 106 |
|
this level. The top-level routine for the numerical |
| 107 |
|
part of the code is in model/src/the_model_main.F. |
| 108 |
|
|
| 109 |
|
|
| 110 |
|
o References |
| 111 |
|
Web sites - HP |
| 112 |
|
for doc Digital |
| 113 |
|
SGI |
| 114 |
|
Sun |
| 115 |
|
Linux threads |
| 116 |
|
CRAY multitasking |
| 117 |
|
PPT notes |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch