Selecting the number of CPUs for the Hg simulation
First, you need to decide whether you will run the MITgcm as a debugging simulation (with 13 CPUs), or as a full science simulation (with 96 CPUs). You can change the number of CPUs for your MITgcm simulation with the setcpus script:
To select a 13 CPU debugging simulation, type the following commands:
cd /n/regal/sunderland_lab/YOUR-USER-NAME/MITgcm_code/
./setcpus 13 hg
Or to select a 96 CPU science simulation, type:
cd /n/regal/sunderland_lab/YOUR-USER-NAME/MITgcm_code/
./setcpus 96 hg
The setcpus script will copy a couple of files into place with appropriate settings for each simulation.
Changing default Hg simulation parameter settings (if necessary)
If you need to override any of the default settings in the various header files (i.e. to increase parameter sizes), you can place updated files into this directory:
MITgcm_code/verification/global_hg_llc90/code
Here is a list of the most important settings:
| File | Modification |
|---|
PROFILES_SIZE.h | Changed parameter NOBSGLOB from 500 to 400000 to allow MITgcm to read certain files in the run/profiles directory |
HG_OPTIONS.h | You can define (or undefine) several C-preprocessor switches to turn activate (or deactivate) various Hg simulation options. - To activate an option, use
#define OPTION-NAME - To deactivate an option, use
#undef OPTION-NAME
| Option | Description | Default status |
|---|
#define METHY_MODE | Turns on methylmercury | ON | #define FOODW_MODE | Turns on Darwin food web model | ON | #define USE_EXFIWR | Use external forcing fields (needed for online run) | ON | #define BC_Mason1996 | Use Mason et al 1996 bioduluation? | ON | #define PART_NO | Turn off partitioining | OFF | #define PART_EQ | Turn on instantaneous equlibrium | ON | #define PART_DI | Turn on release (disassociation) | OFF | #define HG2REFRACTORY | Use a separate tracer for riverine Hg2 | OFF | #define ISHG2LABILE | Is the Hg2 tracer exclusively labile? | OFF | #define f_depo_refrac | Fraction of refractory Hg2 for atmospheric deposition | 0.0 | #define HGPREFRACTORY | Use a separate tracer for riverine HgP | OFF | #define HGPR_IRREDUCIBLE | Is riverine HgP irreducible? | OFF | #define HGPR_SLOWRELEASING | Is riverine HgP slow-releasing? | ON | #define WALSH_FILTER | Activate Walsh & Nittrouer typology filter for riverine particles? | Activated when #define HGPREFRACTORY is ON | #define walsh_depth | disable sedimentation:
0: no disable
1: only river mouth
2: over depth <= 55 m (top 4 levels)
3: over depth <= 185 m (top 8 levels)
4: over river mouth AND depth shallower than 55 m | 0 | #define VARI_RED_POOL | Is Hg2 reducible pool a function of DOC concentrations? | OFF |
|
HG_SIZE.h | You can modify tracer indices and size parameters for the Hg simulation if necessary: | Parameter | Description | Default status |
|---|
iHg0 | Hg0 tracer index | 1 | iHg2 | Hg2 tracer index | 2 | iHgP | HgP tracer index | 3 | iHg2R | Hg2R (riverine Hg2) tracer index) | 4 if #define HG2REFRACTORY is ON
| iHg2P | HgPR (riverine HgP tracer index) | 5 (if #define HG2REFRACTORY and #define HGPREFRACTORY are both ON)
4 (if #define HGPREFRACTORY is ON but #define HG2REFRACTORY is OFF) | npmax | Number of functional groups of phytoplankton | 6 (only if #define FOODW_MODE is ON) | nzmax | Number of functional groups of zooplankton | 2 (only if #define FOODW_MODE is ON) | nrmax | Number of river inputs | 1 (only if #define HGPREFRACTORY is ON) | iMMHg | Methyl mercury tracer index | iHgPR + nrmax (only if #define METHY_MODE and #define HGPREFRACTORY is ON)
iHg2PR + 1 (only if #define_METHY_MODE and #define HG2REFRACTORY is ON) | iMMHgP | Particulate methyl mercury tracer index | iMMHgP + 1 (only if #define METHY_MODE is ON) | iDMHg | Dissolved  Methyl mercury tracer index | iDMHg + 1 |
|
You will probably not need to change any of the above settings unless you are making your own modifications to the Hg simulation.
Building the mitgcmuv executable for the Hg simulation
Once you are sure that you have made the proper settings in MITgcm_code/verification/global_hg_llc90/code directory, change to the build directory:
cd ../build
and type the following commands:
premake
make -j13 depend
make -j13
cp mitgcmuv ../run
Each of these commands will be described in more detail below:
| Command | Description |
|---|
premake | premake is an alias in your ~/.bashrc_mitgcm file.
premake calls the MITgcm_code/tools/genmake program, which
- Queries your computer system to find out where various libraries (i.e. MPI, netCDF, etc) are located,
- Uses the compilation options that you specify in the file
MITgcm_code/tools/build_options/inux_ifort_odyssey - Generates the main MITgcm makefiles.
- For more information about genmake, see the MITgcm User Manual, Chapter 3.4.2.
|
make -j13 depend | This command builds the “dependencies” listing and adds it to the Makefile. In other words, it tells the Makefile which source code files depend on which other files.
This is done in order to reduce compilation time. - NOTE: the
-j13 option tells the Make command to use 13 CPUs.
- This number should be the same as the number of CPUs your interactive session uses. This will reduce the overall compilation time.
|
make -j13 | Builds the MITgcm source code into an executable. - A symbolic link to each Fortran (
*.F) and include file (*.h) will be made in the build/ directory. - For each *.F file, include files will be inlined and C-preprocessor switches evaluated. The resultant "clean" Fortran file will be saved with a *.f extension.
- The *.f files will be compiled into an executable file named mitgcmuv.
|
cp mitgcmuv ../run | Copies the mitgcmuv executable to the run directory, where it will be used to run MITgcm simulations. |
Previous | Up | Next