| 1 |
$Header$ |
$Header$ |
| 2 |
|
|
| 3 |
|
|
| 4 |
MITgcmUV Getting Started |
MITgcmUV Getting Started |
| 5 |
======================== |
======================== |
| 6 |
|
|
| 7 |
o Introduction |
o Introduction |
| 8 |
|
|
| 9 |
o Install |
This note is a guide to using the MIT General Circulation Model Ultra Verstaile |
| 10 |
tar -xvf ... |
implementation, MITgmcUV. MITgcmUV is a Fortran code that implements the |
| 11 |
cd tools |
algorithm described in Marshall et. al. 1997, Hill, Adcroft, ... |
| 12 |
genmake.dec |
The MITgcmUV implementation is designed to work efficiently on all classes |
| 13 |
cd ../bin |
of computer platforms. It can be used in both a single processor mode |
| 14 |
ln -s ../tools/Makefile.dec makefile |
and a parallel processor mode. Parallel processing can be either multi-threaded |
| 15 |
make depend |
shared memory such as that found on CRAY T90 machines or it can be multi-process |
| 16 |
make |
distributed memory. A set of "execution enviroment" support routines are |
| 17 |
setenv PARALLEL 1 |
used to allow the same numerical code to run on top of a single-process, multi-threaded |
| 18 |
../exe/barrier |
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 cNN.tar |
| 28 |
|
|
| 29 |
|
2. Create platform specific make file |
| 30 |
|
For example on a Digital UNIX machine the script "genmake" 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. To run the model with example data see the "README" in each of |
| 44 |
|
the example experiments in ../verification/exp*/ |
| 45 |
|
e.g. ../verification/exp2/README |
| 46 |
|
|
| 47 |
|
|
| 48 |
o Running |
o Running |
| 49 |
|
|
| 50 |
- Inout files |
- Input and output files |
| 51 |
eedata |
|
| 52 |
data |
Required files |
| 53 |
|
============== |
| 54 |
|
The model is configured to look for two files with fixed names. |
| 55 |
|
These files are called |
| 56 |
|
"eedata" and "data". |
| 57 |
|
The file eedata contains "execution environment" data. At present |
| 58 |
|
this consists of a specification of the number of threads to |
| 59 |
|
use in X and Y under multithreaded execution. |
| 60 |
|
|
| 61 |
- Serial execution |
- Serial execution |
| 62 |
|
|
| 63 |
|
- Parallel execution. MPI |
| 64 |
|
mPx, nPy |
| 65 |
|
dmpirun |
| 66 |
|
|
| 67 |
- Parallel execution. Threads |
- Parallel execution. Threads |
| 68 |
nSx, nSy |
nSx, nSy |
| 69 |
setenv PARALLEL n |
setenv PARALLEL n |
| 70 |
nTx=2, nTy=2 |
nTx=2, nTy=2 |
| 71 |
|
|
|
- Parallel execution. MPI |
|
|
mPx, nPy |
|
|
dmpirun |
|
|
|
|
| 72 |
- Parallel execution. Hybrid |
- Parallel execution. Hybrid |
| 73 |
|
|
| 74 |
|
o Cutomising the code |
| 75 |
|
|
| 76 |
|
Model structure |
| 77 |
|
=============== |
| 78 |
|
The "numerical" model is contained within a execution |
| 79 |
|
environment support wrapper. This wrapper is designed |
| 80 |
|
to provide a general framework for grid-point models. |
| 81 |
|
MITgcmUV is a specific numerical model that uses the |
| 82 |
|
framework. |
| 83 |
|
Under this structure the model is split into |
| 84 |
|
execution environment support code and conventional |
| 85 |
|
numerical model code. The execution environment |
| 86 |
|
support code is held under the "eesupp" directory. |
| 87 |
|
The grid point model code is held under the |
| 88 |
|
"model" directory. |
| 89 |
|
Code execution actually starts in the eesupp |
| 90 |
|
routines and not in the model routines. For this |
| 91 |
|
reason the top level main.F is in the eesupp/src |
| 92 |
|
directory. End-users should not need to worry about |
| 93 |
|
this level. The top-level routine for the numerical |
| 94 |
|
part of the code is in model/src/the_model_main.F. |
| 95 |
|
|
| 96 |
|
|
| 97 |
o References |
o References |
| 98 |
Web sites - HP |
Web sites - HP |
| 99 |
for doc Digital |
for doc Digital |
| 101 |
Sun |
Sun |
| 102 |
Linux threads |
Linux threads |
| 103 |
CRAY multitasking |
CRAY multitasking |
| 104 |
|
PPT notes |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch