ecearth3/sources/tm5mp/rc/machine-platform.rc.tmpl

!===============================================
! compiler information
!===============================================
!
! All the keys in this section should be filled here or put in a compiler rc
! file. See the other machine rc files for examples.
! 
! So, either include a file:

#include rc/compiler-<name>-<version>.rc

! *OR* set all these keys:

! Compiler for Fortran 90, F77 (fixed form), openMP, MPI wrappers. It is
! possible that the same compiler is used for all.
compiler.fc             :  
compiler.f77            :  
compiler.fc.openmp      :  
mpi.compiler.fc         :  
mpi.compiler.fc.openmp  :  

! flag to retrieve the version of the compiler
compiler.getversion_flag : 

! *** For all the remaining flags, you must specify one for compilation and
! *** one for linking.

! defaults flags are always used, regardless of optimization and runtime checks
compiler.flags.default.fflags       :  
compiler.flags.default.ldflags      : 

! to enforce all real to be double precision
compiler.flags.real8.fflags         :  
compiler.flags.real8.ldflags        : 

! to use MPI
compiler.flags.mpi.fflags           : 
compiler.flags.mpi.ldflags          : 

! to use openMP
compiler.flags.openmp.fflags        : 
compiler.flags.openmp.ldflags       : 


! ** the pycasso script expects four levels of optimization that you define
! ** here. You have to choose one of them in your main rc file.

compiler.flags.optim-none.fflags    : 
compiler.flags.optim-none.ldflags   : 

compiler.flags.optim-strict.fflags  : 
compiler.flags.optim-strict.ldflags :  

compiler.flags.optim-fast.fflags    : 
compiler.flags.optim-fast.ldflags   : 

compiler.flags.optim-vfast.fflags   : 
compiler.flags.optim-vfast.ldflags  : 

! ** you can also define two sets of runtime check flags, that you can add in
! ** your main rc file when developing
compiler.flags.check-all.fflags     : 
compiler.flags.check-all.ldflags    : 

compiler.flags.debug.fflags         : 
compiler.flags.debug.ldflags        : 


!===============================================
! libraries for HDF4 support (mandatory)
!===============================================

! Z library (used for compression in HDF)
compiler.lib.z.fflags           : 
compiler.lib.z.libs             : 

! JPEG library (used for compression in HDF)
JPEG_HOME                       : 
compiler.lib.jpeg.fflags        : 
compiler.lib.jpeg.libs          : 

! SZ library (used for compression in HDF)
SZIP_HOME                       : 
compiler.lib.sz.fflags          : 
compiler.lib.sz.libs            : 

! HDF4 library (without netcdf interface)
HDF4_HOME                       : 
compiler.lib.hdf4.fflags        : 
compiler.lib.hdf4.libs          : 

!===============================================
! libraries for NetCDF support (mandatory)
!===============================================
!
! Use one netcdf library, optionaly with hdf5 if needed. To do, so add the
! corresponding 'with_*' to the 'my.df.define' key in your main rc file.

! NetCDF3 library (with_netcdf)
NETCDF_HOME                     : 
compiler.lib.netcdf.fflags      : 
compiler.lib.netcdf.libs        : 

! NetCDF4 library (with_netdf4)
NETCDF4_HOME                    : 
compiler.lib.netcdf4.fflags     : 
compiler.lib.netcdf4.libs       : 

! NetCDF4 library with parallel IO features (with_netcdf4_par)
NETCDF4_MPI_HOME                : 
compiler.lib.netcdf4_par.fflags : 
compiler.lib.netcdf4_par.libs   : 

! * since NetCDF4 is build on top of HDF5, you may have to link/include HDF5
! * library when building the code:

! HDF5 library (with_hdf5)
HDF5_HOME                       : 
compiler.lib.hdf5.fflags        : 
compiler.lib.hdf5.libs          : 

! HDF5 library with parallel features (with_hdf5_par)
HDF5_MPI_HOME                   : 
compiler.lib.hdf5_par.fflags    : 
compiler.lib.hdf5_par.libs      : 

!===============================================
! Other libraries
!===============================================

! MPI library
MPI_HOME                        : 
compiler.lib.mpi.fflags         : 
compiler.lib.mpi.libs           : 


! GRIB-API library (optional). It is needed only if you read meteo files from
! the ECMWF MARS system, where they are in grib format.
compiler.lib.grib_api.fflags : 
compiler.lib.grib_api.libs   : 

! a UDUNITS library is optional. It can be used if you are reading meteo files
! in netCDF format to check that units conversion are correct. It is known to
! slow down the code significantly. The two main version are supported.

! Udunits v1.x library (add with_udunits1 to my.tmm.define in your main rc file)
UDUNITS_HOME                    :  
compiler.lib.udunits1.fflags    : 
compiler.lib.udunits1.libs      : 

! Udunits v2.x library (add with_udunits2 to my.tmm.define in your main rc file)
compiler.lib.udunits2.fflags    :  
compiler.lib.udunits2.libs      :  

! LAPACK library (optional). If available, add the with_lapack to
! my.tm5.define in your main rc file. see User Manual for further details.
LAPACK_HOME                     : 
compiler.lib.lapack.fflags      : 
compiler.lib.lapack.libs        : 


!===============================================================
! SETTINGS FOR BATCH SCHEDULER (qsub,bsub,loadleveler,slurm,pbs)
!===============================================================
!
! Settings in this section can be in a dedicated rc file, or just left here. So,
! either include such a file:

#include  rc/queue-<name>-<institute>.rc

! *OR* fill the following keys:

queue : <name-of-your-job-manager>

! arguments passed directly to submit command
queue.pbs.submit.options :

! number of MPI tasks (do not change)
my.queue.ntask : ${par.ntask}

! number of threads (always 1, openMP not supported in TM5-MP)
my.queue.nthread : 1


! Queue options are set below, and managed by the 'bin/submit_tm5_tools.py'
! script. There can be as many as needed.
!
! Some special values can be used:
!    if set to <none>, the option is skipped
!    if set to <auto>, it is automatically set in the python script
!    if empty,         it is used as 'option without argument'. 
!                    
! But check that they are implemented for you job manager. See
! bin/submit_tm5_tools.py/QueueOptions_<scheduler-name> for details. 
!
! If needed, user account is available from the main rc file as ${my.queue.account}

!-----------------------------------------------
! ** queue options with same value for ALL steps
!-----------------------------------------------
! space seperated list
queue.<scheduler-name>.options.all : <option1> <option2> <option3>...
! and their values
queue.<scheduler-name>.option.all.<option1> : 
queue.<scheduler-name>.option.all.<option2> : 
queue.<scheduler-name>.option.all.<option3> : 

!------------------------------------------------
! ** queue options with different values per step
!------------------------------------------------
! list options per step:
queue.<scheduler-name>.options.init : <ini_option1> <ini_option2> <ini_option3>...
queue.<scheduler-name>.options.run  : <run_option1> <run_option2>...
queue.<scheduler-name>.options.done : <don_option1> <don_option2>...

! ~~~~~ step init options values
queue.<scheduler-name>.option.init.<ini_option1> : 
queue.<scheduler-name>.option.init.<ini_option2> : 
queue.<scheduler-name>.option.init.<ini_option3> : 

! ~~~~~ step run options values
queue.<scheduler-name>.option.run.<run_option1> : 
queue.<scheduler-name>.option.run.<run_option2> : 

! ~~~~~ step done options values
queue.<scheduler-name>.option.done.<don_option1> : 
queue.<scheduler-name>.option.done.<don_option2> : 


!===============================================
! make
!===============================================
!
! Specify the make command. The setup script will insert the 'build.jobs'
! specified in the expert.rc (probably 8). But can be overwritten at the
! command line (argument to 'setup_tm5' script)
! 
maker :  gmake -j %{build.jobs}

!JOB-COMPILE  ! If you want to submit the compilation to a compute node, uncomment and set
!JOB-COMPILE  ! the necessary keys. See user manual for the further details.
!JOB-COMPILE  ! 
!JOB-COMPILE  ! - if F (default), pycasso calls the ${maker} command in the foreground.  
!JOB-COMPILE  ! - if T, then compilation is submitted according to "submit.to" key (default
!JOB-COMPILE  !         of which is set in expert.rc, and can be overwritten at the CLI): either "queue", or foreground.
!JOB-COMPILE  !
!JOB-COMPILE  ! "my.queue.make" should be set in the main rc file, so the user can switch it on/off from there.
!JOB-COMPILE  ! 
!JOB-COMPILE  build.make.submit   : ${my.queue.make}
!JOB-COMPILE   
!JOB-COMPILE  ! list of options for the job manager, and their value
!JOB-COMPILE  queue.<scheduler-name>.options.build  : option1 option2 option3...
!JOB-COMPILE  queue.<scheduler-name>.option.build.<option1> : 
!JOB-COMPILE  queue.<scheduler-name>.option.build.<option2> : 
!JOB-COMPILE  queue.<scheduler-name>.option.build.<option3> : 


!===============================================
! MPI runner
!===============================================
!
! command for running MPI parallel programs

mpirun.command     :  
mpirun.args        :  

! name of command file; if empty, then executable and arguments are added to the command line
mpirun.cmdfile     :  

! name of host file
mpirun.hostfile    :  


!===============================================
! debugger
!===============================================
!
! type: totalview | idb | kdbg
debugger                :  kdbg

! command for debugger
! o KDE debugger around gdb (Linux systems using gfortran)
debugger.command        :  kdbg
! o Intel debugger (for systems with Intel compiler)
!debugger.command        :  idb -gui
! o TotalView (IBM)
!debugger.command        :  totalview -searchPath=${build.sourcedir}


!===============================================
! model data
!===============================================
! the user scratch directory
my.scratch          :  ${SCRATCH}

! *Permanent* archives to search for meteo files
! (Note that the location of the meteo files at runtime is set in the main rc)
my.meteo.search     :  

! base path to other input data files distributed with the model
my.data.dir         :  

! extra install tasks
my.install.tasks    :