Changes between Version 64 and Version 65 of Doc/FAQ
- Timestamp:
- 11/08/19 17:28:25 (4 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
Doc/FAQ
v64 v65 68 68 ## How do I define the number of MPI jobs and the number of OpenMP threads? ## 69 69 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. 70 It's define in config.card file, and the script ins_job will define what it's need in the job header. 71 [[BR]] 72 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". 73 For example, for LMDZOR : we choose to run with 71 MPI processes and 8 OpenMP threads for LMDZ, and 1 MPI for XIOS 74 {{{ 75 ATM= (gcm.e, lmdz.x, 71MPI, 8OMP) 76 SRF= ("", "") 77 SBG= ("", "") 78 IOS= (xios_server.exe, xios.x, 1MPI) 79 }}} 80 In this case the job will ask for 71*8 +1 = 569 CPU 81 [[BR]] 82 If we don't use OpenMP parallelization 83 {{{ 84 ATM= (gcm.e, lmdz.x, 71MPI, 1OMP) 85 SRF= ("", "") 86 SBG= ("", "") 87 IOS= (xios_server.exe, xios.x, 1MPI) 88 }}} 89 In this case the job will ask for 71 +1 = 72 CPU 90 84 91 85 92 … … 108 115 cd modipsl 109 116 mv 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 }}} 117 svn checkout -r `number_revision` http://forge.ipsl.jussieu.fr/libigcm/svn/trunk/libIGCM libIGCM 118 }}} 119 where number_revision is specified by someone from PlatForm group. 119 120 120 121 … … 132 133 ## How do I restart a simulation to recover missing output files? ## 133 134 134 TO BE DONE ON JEANZAY (idea : change $STOREDIR by jeanzay workdir)135 136 135 This method shows how to rerun a complete simulation period in a different directory (REDO instead of DEVT/PROD). 137 136 137 For reminder 138 || Space || TGCC || IDRIS || 139 || WORK || $CCCWORKDIR || $WORK || 140 || SCRATCH || $CCCSCRATCHDIR || $SCRATCH || 141 || STORE || $CCCSTOREDIR || $STORE || 142 143 138 144 Example : 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. 140 146 {{{ 141 147 ## Directory 142 mkdir $CCCSTOREDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1143 cd $CCCSTOREDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1148 mkdir STORE/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1 149 cd STORE/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1 144 150 # RESTART 145 151 mkdir -p RESTART ; cd RESTART … … 149 155 * If you are running a coupled model : On the scratch disk ($CCCSCRATCHDIR/IGCM_OUT), create the mesh_mask file 150 156 {{{ 151 mkdir $CCCSCRATCHDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO152 cd $CCCSCRATCHDIR/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO157 mkdir SCRATCH/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO 158 cd SCRATCH/....IGCM_OUT/IPSLCM5A/REDO/historicalAnt/v3.historicalAnt1REDO 153 159 # mesh_mask 154 160 mkdir -p OCE/Output … … 273 279 [BoundaryFiles] 274 280 List= ....\ 275 ( /ccc/work/cont003/subipsl/subipsl/ECMWF96x95/AN${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)\ 277 283 }}} 278 284 * choose the proper dates in `config.card` (pay attention to leap years) … … 467 473 468 474 ## How do I check that the post processing jobs were successful? ## 469 470 See [wiki:Doc/Running#Howtocheckthatthepostprocessingwassuccessful here]. 475 see [wiki:Doc/CheckDebug#Checkstatusofyoursimulations here] 471 476 472 477 ## How do I read/retrieve/use files on esgf/thredds? ## … … 524 529 525 530 ## What is the Monitoring? ## 526 See chapter ''' Simulation and post-processing''''''', section ''Postprocessing with libIGCM'' [wiki:Doc/Running#Monitoringandintermonitoring here]531 See chapter '''Run and post-proc''''''', section ''Monitoring and Intermonitoring'' [wiki:Doc/Running#Monitoringandintermonitoring here] 527 532 528 533 ## How do I add a plot to the monitoring? ##