Changes between Version 64 and Version 65 of Doc/FAQ


Ignore:
Timestamp:
11/08/19 17:28:25 (4 years ago)
Author:
acosce
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • Doc/FAQ

    v64 v65  
    6868## How do I define the number of MPI jobs and the number of OpenMP threads? ## 
    6969 
    70 If you run your model in MPI mode only (without OpenMP) the number of MPI processes is defined in config.card by the !JobNumProcTot parameter:   
    71 {{{ 
    72 #-- Total Number of Processors 
    73 JobNumProcTot=32 
    74 }}} 
    75  
    76 If you run your model in hybrid mode (MPI-OpenMP), the number of MPI processes and the number of OpenMP threads are set in config.card in the section "Executable". You still need to precise the CPU total number in `JobNumProcTot` parameter by summing all components processes.[[br]][[br]] 
    77  
    78 For instance, for LMDZ : 16 MPI processes and 2 OpenMP threads (ATM compenent needs `2OMP x 16MPI = 32CPU`). 
    79 {{{ 
    80 ATM= (gcm.e, lmdz.x, 16MPI, 2OMP) 
    81 }}} 
    82  
    83 Notice that the job header differs from the one for openMP because you fill the total of core to use. 
     70It's define in config.card file, and the script ins_job will define what it's need in the job header.  
     71[[BR]] 
     72If you run your model in hybrid mode (MPI-OpenMP), the number of MPI processes and the number of OpenMP threads are set in config.card in the section "Executable".  
     73For example, for LMDZOR : we choose to run with 71 MPI processes and 8 OpenMP threads for LMDZ, and 1 MPI for XIOS  
     74{{{ 
     75ATM= (gcm.e, lmdz.x, 71MPI, 8OMP) 
     76SRF= ("", "") 
     77SBG= ("", "") 
     78IOS= (xios_server.exe, xios.x, 1MPI) 
     79}}} 
     80In this case the job will ask for 71*8 +1 = 569 CPU  
     81[[BR]] 
     82If we don't use OpenMP parallelization  
     83{{{ 
     84ATM= (gcm.e, lmdz.x, 71MPI, 1OMP) 
     85SRF= ("", "") 
     86SBG= ("", "") 
     87IOS= (xios_server.exe, xios.x, 1MPI) 
     88}}} 
     89In this case the job will ask for 71 +1 = 72 CPU 
     90 
    8491 
    8592  
     
    108115cd modipsl 
    109116mv libIGCM libIGCM_old 
    110 svn checkout http://forge.ipsl.jussieu.fr/libigcm/svn/tags/libIGCM_v2.8.4 libIGCM 
    111 libIGCM/ins_job # 
    112 }}} 
    113  
    114 In case you need version XXX of the trunk of libIGCM, change the "svn checkout" line into: 
    115  
    116 {{{#!sh 
    117 svn checkout -r XXX http://forge.ipsl.jussieu.fr/libigcm/svn/trunk/libIGCM libIGCM 
    118 }}} 
     117svn checkout -r `number_revision` http://forge.ipsl.jussieu.fr/libigcm/svn/trunk/libIGCM libIGCM 
     118}}} 
     119where number_revision is specified by someone from PlatForm group.  
    119120 
    120121 
     
    132133## How do I restart a simulation to recover missing output files?  ## 
    133134 
    134 TO BE DONE ON JEANZAY (idea : change $STOREDIR by jeanzay workdir)  
    135  
    136135This method shows how to rerun a complete simulation period in a different directory (REDO instead of DEVT/PROD). 
    137136 
     137For reminder  
     138|| Space || TGCC || IDRIS || 
     139|| WORK || $CCCWORKDIR || $WORK || 
     140|| SCRATCH || $CCCSCRATCHDIR || $SCRATCH || 
     141|| STORE || $CCCSTOREDIR || $STORE || 
     142 
     143 
    138144Example : To rerun v3.historicalAnt1 to recompute a whole year (e.g. 1964) you must : 
    139  * On the file server (CCCSTOREDIR), create the necessary RESTART file. 
     145 * On the file server (STORE), create the necessary RESTART file. 
    140146{{{ 
    141147## Directory 
    142 mkdir $CCCSTOREDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1 
    143 cd $CCCSTOREDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1 
     148mkdir STORE/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1 
     149cd STORE/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1 
    144150# RESTART 
    145151mkdir -p RESTART ; cd RESTART 
     
    149155 * If you are running a coupled model : On the scratch disk ($CCCSCRATCHDIR/IGCM_OUT), create the mesh_mask file 
    150156{{{ 
    151 mkdir $CCCSCRATCHDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO 
    152 cd $CCCSCRATCHDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO 
     157mkdir SCRATCH/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO 
     158cd SCRATCH/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO 
    153159# mesh_mask 
    154160mkdir -p OCE/Output 
     
    273279[BoundaryFiles] 
    274280List= ....\ 
    275       (/ccc/work/cont003/subipsl/subipsl/ECMWF96x95/AN${year}/u_ecmwf_${year}${month}.nc, u.nc)\ 
    276       (/ccc/work/cont003/subipsl/subipsl/ECMWF96x95/AN${year}/v_ecmwf_${year}${month}.nc, v.nc)\ 
     281      (work_subipsl/subipsl/ECMWF{your_resolution}/AN${year}/u_ecmwf_${year}${month}.nc, u.nc)\ 
     282      (work_subipsl/subipsl/ECMWF{your_resolution}/AN${year}/v_ecmwf_${year}${month}.nc, v.nc)\ 
    277283}}} 
    278284 * choose the proper dates in `config.card` (pay attention to leap years) 
     
    467473 
    468474## How do I check that the post processing jobs were successful? ## 
    469  
    470 See [wiki:Doc/Running#Howtocheckthatthepostprocessingwassuccessful here]. 
     475see [wiki:Doc/CheckDebug#Checkstatusofyoursimulations here] 
    471476 
    472477## How do I read/retrieve/use files on esgf/thredds? ## 
     
    524529 
    525530## What is the Monitoring? ## 
    526 See chapter '''Simulation and post-processing''''''', section ''Postprocessing with libIGCM''  [wiki:Doc/Running#Monitoringandintermonitoring here] 
     531See chapter '''Run and post-proc''''''', section ''Monitoring and Intermonitoring''  [wiki:Doc/Running#Monitoringandintermonitoring here] 
    527532 
    528533## How do I add a plot to the monitoring? ##