| 1 | MODULE mppini |
|---|
| 2 | !!====================================================================== |
|---|
| 3 | !! *** MODULE mppini *** |
|---|
| 4 | !! Ocean initialization : distributed memory computing initialization |
|---|
| 5 | !!====================================================================== |
|---|
| 6 | !! History : 6.0 ! 1994-11 (M. Guyon) Original code |
|---|
| 7 | !! OPA 7.0 ! 1995-04 (J. Escobar, M. Imbard) |
|---|
| 8 | !! 8.0 ! 1998-05 (M. Imbard, J. Escobar, L. Colombet ) SHMEM and MPI versions |
|---|
| 9 | !! NEMO 1.0 ! 2004-01 (G. Madec, J.M Molines) F90 : free form , north fold jpni > 1 |
|---|
| 10 | !! 3.4 ! 2011-10 (A. C. Coward, NOCS & J. Donners, PRACE) add init_nfdcom |
|---|
| 11 | !! 3. ! 2013-06 (I. Epicoco, S. Mocavero, CMCC) init_nfdcom: setup avoiding MPI communication |
|---|
| 12 | !! 4.0 ! 2016-06 (G. Madec) use domain configuration file instead of bathymetry file |
|---|
| 13 | !! 4.0 ! 2017-06 (J.M. Molines, T. Lovato) merge of mppini and mppini_2 |
|---|
| 14 | !!---------------------------------------------------------------------- |
|---|
| 15 | |
|---|
| 16 | !!---------------------------------------------------------------------- |
|---|
| 17 | !! mpp_init : Lay out the global domain over processors with/without land processor elimination |
|---|
| 18 | !! init_ioipsl: IOIPSL initialization in mpp |
|---|
| 19 | !! init_nfdcom: Setup for north fold exchanges with explicit point-to-point messaging |
|---|
| 20 | !! init_doloop: set the starting/ending indices of DO-loop used in do_loop_substitute |
|---|
| 21 | !!---------------------------------------------------------------------- |
|---|
| 22 | USE dom_oce ! ocean space and time domain |
|---|
| 23 | USE bdy_oce ! open BounDarY |
|---|
| 24 | ! |
|---|
| 25 | USE lbcnfd , ONLY : isendto, nsndto ! Setup of north fold exchanges |
|---|
| 26 | USE lib_mpp ! distribued memory computing library |
|---|
| 27 | USE iom ! nemo I/O library |
|---|
| 28 | USE ioipsl ! I/O IPSL library |
|---|
| 29 | USE in_out_manager ! I/O Manager |
|---|
| 30 | |
|---|
| 31 | IMPLICIT NONE |
|---|
| 32 | PRIVATE |
|---|
| 33 | |
|---|
| 34 | PUBLIC mpp_init ! called by opa.F90 |
|---|
| 35 | |
|---|
| 36 | INTEGER :: numbot = -1 ! 'bottom_level' local logical unit |
|---|
| 37 | INTEGER :: numbdy = -1 ! 'bdy_msk' local logical unit |
|---|
| 38 | |
|---|
| 39 | !!---------------------------------------------------------------------- |
|---|
| 40 | !! NEMO/OCE 4.0 , NEMO Consortium (2018) |
|---|
| 41 | !! $Id$ |
|---|
| 42 | !! Software governed by the CeCILL license (see ./LICENSE) |
|---|
| 43 | !!---------------------------------------------------------------------- |
|---|
| 44 | CONTAINS |
|---|
| 45 | |
|---|
| 46 | #if ! defined key_mpp_mpi |
|---|
| 47 | !!---------------------------------------------------------------------- |
|---|
| 48 | !! Default option : shared memory computing |
|---|
| 49 | !!---------------------------------------------------------------------- |
|---|
| 50 | |
|---|
| 51 | SUBROUTINE mpp_init |
|---|
| 52 | !!---------------------------------------------------------------------- |
|---|
| 53 | !! *** ROUTINE mpp_init *** |
|---|
| 54 | !! |
|---|
| 55 | !! ** Purpose : Lay out the global domain over processors. |
|---|
| 56 | !! |
|---|
| 57 | !! ** Method : Shared memory computing, set the local processor |
|---|
| 58 | !! variables to the value of the global domain |
|---|
| 59 | !!---------------------------------------------------------------------- |
|---|
| 60 | ! |
|---|
| 61 | jpiglo = Ni0glo |
|---|
| 62 | jpjglo = Nj0glo |
|---|
| 63 | jpimax = jpiglo |
|---|
| 64 | jpjmax = jpjglo |
|---|
| 65 | jpi = jpiglo |
|---|
| 66 | jpj = jpjglo |
|---|
| 67 | jpk = jpkglo |
|---|
| 68 | jpim1 = jpi-1 ! inner domain indices |
|---|
| 69 | jpjm1 = jpj-1 ! " " |
|---|
| 70 | jpkm1 = MAX( 1, jpk-1 ) ! " " |
|---|
| 71 | ! |
|---|
| 72 | CALL init_doloop ! set start/end indices or do-loop depending on the halo width value (nn_hls) |
|---|
| 73 | ! |
|---|
| 74 | jpij = jpi*jpj |
|---|
| 75 | jpni = 1 |
|---|
| 76 | jpnj = 1 |
|---|
| 77 | jpnij = jpni*jpnj |
|---|
| 78 | nimpp = 1 ! |
|---|
| 79 | njmpp = 1 |
|---|
| 80 | nbondi = 2 |
|---|
| 81 | nbondj = 2 |
|---|
| 82 | nidom = FLIO_DOM_NONE |
|---|
| 83 | npolj = 0 |
|---|
| 84 | IF( jperio == 3 .OR. jperio == 4 ) npolj = 3 |
|---|
| 85 | IF( jperio == 5 .OR. jperio == 6 ) npolj = 5 |
|---|
| 86 | l_Iperio = jpni == 1 .AND. (jperio == 1 .OR. jperio == 4 .OR. jperio == 6 .OR. jperio == 7) |
|---|
| 87 | l_Jperio = jpnj == 1 .AND. (jperio == 2 .OR. jperio == 7) |
|---|
| 88 | ! |
|---|
| 89 | IF(lwp) THEN |
|---|
| 90 | WRITE(numout,*) |
|---|
| 91 | WRITE(numout,*) 'mpp_init : NO massively parallel processing' |
|---|
| 92 | WRITE(numout,*) '~~~~~~~~ ' |
|---|
| 93 | WRITE(numout,*) ' l_Iperio = ', l_Iperio, ' l_Jperio = ', l_Jperio |
|---|
| 94 | WRITE(numout,*) ' npolj = ', npolj , ' njmpp = ', njmpp |
|---|
| 95 | ENDIF |
|---|
| 96 | ! |
|---|
| 97 | IF( jpni /= 1 .OR. jpnj /= 1 .OR. jpnij /= 1 ) & |
|---|
| 98 | CALL ctl_stop( 'mpp_init: equality jpni = jpnj = jpnij = 1 is not satisfied', & |
|---|
| 99 | & 'the domain is lay out for distributed memory computing!' ) |
|---|
| 100 | ! |
|---|
| 101 | END SUBROUTINE mpp_init |
|---|
| 102 | |
|---|
| 103 | #else |
|---|
| 104 | !!---------------------------------------------------------------------- |
|---|
| 105 | !! 'key_mpp_mpi' MPI massively parallel processing |
|---|
| 106 | !!---------------------------------------------------------------------- |
|---|
| 107 | |
|---|
| 108 | |
|---|
| 109 | SUBROUTINE mpp_init |
|---|
| 110 | !!---------------------------------------------------------------------- |
|---|
| 111 | !! *** ROUTINE mpp_init *** |
|---|
| 112 | !! |
|---|
| 113 | !! ** Purpose : Lay out the global domain over processors. |
|---|
| 114 | !! If land processors are to be eliminated, this program requires the |
|---|
| 115 | !! presence of the domain configuration file. Land processors elimination |
|---|
| 116 | !! is performed if jpni x jpnj /= jpnij. In this case, using the MPP_PREP |
|---|
| 117 | !! preprocessing tool, help for defining the best cutting out. |
|---|
| 118 | !! |
|---|
| 119 | !! ** Method : Global domain is distributed in smaller local domains. |
|---|
| 120 | !! Periodic condition is a function of the local domain position |
|---|
| 121 | !! (global boundary or neighbouring domain) and of the global |
|---|
| 122 | !! periodic |
|---|
| 123 | !! Type : jperio global periodic condition |
|---|
| 124 | !! |
|---|
| 125 | !! ** Action : - set domain parameters |
|---|
| 126 | !! nimpp : longitudinal index |
|---|
| 127 | !! njmpp : latitudinal index |
|---|
| 128 | !! narea : number for local area |
|---|
| 129 | !! nbondi : mark for "east-west local boundary" |
|---|
| 130 | !! nbondj : mark for "north-south local boundary" |
|---|
| 131 | !! nproc : number for local processor |
|---|
| 132 | !! noea : number for local neighboring processor |
|---|
| 133 | !! nowe : number for local neighboring processor |
|---|
| 134 | !! noso : number for local neighboring processor |
|---|
| 135 | !! nono : number for local neighboring processor |
|---|
| 136 | !!---------------------------------------------------------------------- |
|---|
| 137 | INTEGER :: ji, jj, jn, jproc, jarea ! dummy loop indices |
|---|
| 138 | INTEGER :: inijmin |
|---|
| 139 | INTEGER :: i2add |
|---|
| 140 | INTEGER :: inum ! local logical unit |
|---|
| 141 | INTEGER :: idir, ifreq, icont ! local integers |
|---|
| 142 | INTEGER :: ii, il1, ili, imil ! - - |
|---|
| 143 | INTEGER :: ij, il2, ilj, ijm1 ! - - |
|---|
| 144 | INTEGER :: iino, ijno, iiso, ijso ! - - |
|---|
| 145 | INTEGER :: iiea, ijea, iiwe, ijwe ! - - |
|---|
| 146 | INTEGER :: iarea0 ! - - |
|---|
| 147 | INTEGER :: ierr, ios ! |
|---|
| 148 | INTEGER :: inbi, inbj, iimax, ijmax, icnt1, icnt2 |
|---|
| 149 | LOGICAL :: llbest, llauto |
|---|
| 150 | LOGICAL :: llwrtlay |
|---|
| 151 | LOGICAL :: ln_listonly |
|---|
| 152 | INTEGER, ALLOCATABLE, DIMENSION(:) :: iin, ii_nono, ii_noea ! 1D workspace |
|---|
| 153 | INTEGER, ALLOCATABLE, DIMENSION(:) :: ijn, ii_noso, ii_nowe ! - - |
|---|
| 154 | INTEGER, ALLOCATABLE, DIMENSION(:,:) :: iimppt, ijpi, ibondi, ipproc ! 2D workspace |
|---|
| 155 | INTEGER, ALLOCATABLE, DIMENSION(:,:) :: ijmppt, ijpj, ibondj, ipolj ! - - |
|---|
| 156 | INTEGER, ALLOCATABLE, DIMENSION(:,:) :: iie0, iis0, iono, ioea ! - - |
|---|
| 157 | INTEGER, ALLOCATABLE, DIMENSION(:,:) :: ije0, ijs0, ioso, iowe ! - - |
|---|
| 158 | LOGICAL, ALLOCATABLE, DIMENSION(:,:) :: llisoce ! - - |
|---|
| 159 | NAMELIST/nambdy/ ln_bdy, nb_bdy, ln_coords_file, cn_coords_file, & |
|---|
| 160 | & ln_mask_file, cn_mask_file, cn_dyn2d, nn_dyn2d_dta, & |
|---|
| 161 | & cn_dyn3d, nn_dyn3d_dta, cn_tra, nn_tra_dta, & |
|---|
| 162 | & ln_tra_dmp, ln_dyn3d_dmp, rn_time_dmp, rn_time_dmp_out, & |
|---|
| 163 | & cn_ice, nn_ice_dta, & |
|---|
| 164 | & ln_vol, nn_volctl, nn_rimwidth |
|---|
| 165 | NAMELIST/nammpp/ jpni, jpnj, ln_nnogather, ln_listonly |
|---|
| 166 | !!---------------------------------------------------------------------- |
|---|
| 167 | ! |
|---|
| 168 | llwrtlay = lwm .OR. sn_cfctl%l_layout |
|---|
| 169 | ! |
|---|
| 170 | ! 0. read namelists parameters |
|---|
| 171 | ! ----------------------------------- |
|---|
| 172 | ! |
|---|
| 173 | READ ( numnam_ref, nammpp, IOSTAT = ios, ERR = 901 ) |
|---|
| 174 | 901 IF( ios /= 0 ) CALL ctl_nam ( ios , 'nammpp in reference namelist' ) |
|---|
| 175 | READ ( numnam_cfg, nammpp, IOSTAT = ios, ERR = 902 ) |
|---|
| 176 | 902 IF( ios > 0 ) CALL ctl_nam ( ios , 'nammpp in configuration namelist' ) |
|---|
| 177 | ! |
|---|
| 178 | IF(lwp) THEN |
|---|
| 179 | WRITE(numout,*) ' Namelist nammpp' |
|---|
| 180 | IF( jpni < 1 .OR. jpnj < 1 ) THEN |
|---|
| 181 | WRITE(numout,*) ' jpni and jpnj will be calculated automatically' |
|---|
| 182 | ELSE |
|---|
| 183 | WRITE(numout,*) ' processor grid extent in i jpni = ', jpni |
|---|
| 184 | WRITE(numout,*) ' processor grid extent in j jpnj = ', jpnj |
|---|
| 185 | ENDIF |
|---|
| 186 | WRITE(numout,*) ' avoid use of mpi_allgather at the north fold ln_nnogather = ', ln_nnogather |
|---|
| 187 | ENDIF |
|---|
| 188 | ! |
|---|
| 189 | IF(lwm) WRITE( numond, nammpp ) |
|---|
| 190 | ! |
|---|
| 191 | !!!------------------------------------ |
|---|
| 192 | !!! nn_hls shloud be read in nammpp |
|---|
| 193 | !!!------------------------------------ |
|---|
| 194 | jpiglo = Ni0glo + 2 * nn_hls |
|---|
| 195 | jpjglo = Nj0glo + 2 * nn_hls |
|---|
| 196 | ! |
|---|
| 197 | ! do we need to take into account bdy_msk? |
|---|
| 198 | READ ( numnam_ref, nambdy, IOSTAT = ios, ERR = 903) |
|---|
| 199 | 903 IF( ios /= 0 ) CALL ctl_nam ( ios , 'nambdy in reference namelist (mppini)' ) |
|---|
| 200 | READ ( numnam_cfg, nambdy, IOSTAT = ios, ERR = 904 ) |
|---|
| 201 | 904 IF( ios > 0 ) CALL ctl_nam ( ios , 'nambdy in configuration namelist (mppini)' ) |
|---|
| 202 | ! |
|---|
| 203 | IF( ln_read_cfg ) CALL iom_open( cn_domcfg, numbot ) |
|---|
| 204 | IF( ln_bdy .AND. ln_mask_file ) CALL iom_open( cn_mask_file, numbdy ) |
|---|
| 205 | ! |
|---|
| 206 | IF( ln_listonly ) CALL bestpartition( MAX(mppsize,jpni*jpnj), ldlist = .TRUE. ) ! must be done by all core |
|---|
| 207 | ! |
|---|
| 208 | ! 1. Dimension arrays for subdomains |
|---|
| 209 | ! ----------------------------------- |
|---|
| 210 | ! |
|---|
| 211 | ! If dimensions of processors grid weren't specified in the namelist file |
|---|
| 212 | ! then we calculate them here now that we have our communicator size |
|---|
| 213 | IF(lwp) THEN |
|---|
| 214 | WRITE(numout,*) 'mpp_init:' |
|---|
| 215 | WRITE(numout,*) '~~~~~~~~ ' |
|---|
| 216 | WRITE(numout,*) |
|---|
| 217 | ENDIF |
|---|
| 218 | IF( jpni < 1 .OR. jpnj < 1 ) THEN |
|---|
| 219 | CALL bestpartition( mppsize, jpni, jpnj ) ! best mpi decomposition for mppsize mpi processes |
|---|
| 220 | llauto = .TRUE. |
|---|
| 221 | llbest = .TRUE. |
|---|
| 222 | ELSE |
|---|
| 223 | llauto = .FALSE. |
|---|
| 224 | CALL bestpartition( mppsize, inbi, inbj, icnt2 ) ! best mpi decomposition for mppsize mpi processes |
|---|
| 225 | ! largest subdomain size for mpi decoposition jpni*jpnj given in the namelist |
|---|
| 226 | CALL basic_decomposition( jpiglo, jpjglo, nn_hls, jpni, jpnj, jpimax, jpjmax ) |
|---|
| 227 | ! largest subdomain size for mpi decoposition inbi*inbj given by bestpartition |
|---|
| 228 | CALL basic_decomposition( jpiglo, jpjglo, nn_hls, inbi, inbj, iimax, ijmax ) |
|---|
| 229 | icnt1 = jpni*jpnj - mppsize ! number of land subdomains that should be removed to use mppsize mpi processes |
|---|
| 230 | IF(lwp) THEN |
|---|
| 231 | WRITE(numout,9000) ' The chosen domain decomposition ', jpni, ' x ', jpnj, ' with ', icnt1, ' land subdomains' |
|---|
| 232 | WRITE(numout,9002) ' - uses a total of ', mppsize,' mpi process' |
|---|
| 233 | WRITE(numout,9000) ' - has mpi subdomains with a maximum size of (jpi = ', jpimax, ', jpj = ', jpjmax, & |
|---|
| 234 | & ', jpi*jpj = ', jpimax*jpjmax, ')' |
|---|
| 235 | WRITE(numout,9000) ' The best domain decompostion ', inbi, ' x ', inbj, ' with ', icnt2, ' land subdomains' |
|---|
| 236 | WRITE(numout,9002) ' - uses a total of ', inbi*inbj-icnt2,' mpi process' |
|---|
| 237 | WRITE(numout,9000) ' - has mpi subdomains with a maximum size of (jpi = ', iimax, ', jpj = ', ijmax, & |
|---|
| 238 | & ', jpi*jpj = ', iimax* ijmax, ')' |
|---|
| 239 | ENDIF |
|---|
| 240 | IF( iimax*ijmax < jpimax*jpjmax ) THEN ! chosen subdomain size is larger that the best subdomain size |
|---|
| 241 | llbest = .FALSE. |
|---|
| 242 | IF ( inbi*inbj-icnt2 < mppsize ) THEN |
|---|
| 243 | WRITE(ctmp1,*) ' ==> You could therefore have smaller mpi subdomains with less mpi processes' |
|---|
| 244 | ELSE |
|---|
| 245 | WRITE(ctmp1,*) ' ==> You could therefore have smaller mpi subdomains with the same number of mpi processes' |
|---|
| 246 | ENDIF |
|---|
| 247 | CALL ctl_warn( ' ', ctmp1, ' ', ' --- YOU ARE WASTING CPU... ---', ' ' ) |
|---|
| 248 | ELSE IF ( iimax*ijmax == jpimax*jpjmax .AND. (inbi*inbj-icnt2) < mppsize) THEN |
|---|
| 249 | llbest = .FALSE. |
|---|
| 250 | WRITE(ctmp1,*) ' ==> You could therefore have the same mpi subdomains size with less mpi processes' |
|---|
| 251 | CALL ctl_warn( ' ', ctmp1, ' ', ' --- YOU ARE WASTING CPU... ---', ' ' ) |
|---|
| 252 | ELSE |
|---|
| 253 | llbest = .TRUE. |
|---|
| 254 | ENDIF |
|---|
| 255 | ENDIF |
|---|
| 256 | |
|---|
| 257 | ! look for land mpi subdomains... |
|---|
| 258 | ALLOCATE( llisoce(jpni,jpnj) ) |
|---|
| 259 | CALL is_ocean( jpni, jpnj, llisoce ) |
|---|
| 260 | inijmin = COUNT( llisoce ) ! number of oce subdomains |
|---|
| 261 | |
|---|
| 262 | IF( mppsize < inijmin ) THEN ! too many oce subdomains: can happen only if jpni and jpnj are prescribed... |
|---|
| 263 | WRITE(ctmp1,9001) ' With this specified domain decomposition: jpni = ', jpni, ' jpnj = ', jpnj |
|---|
| 264 | WRITE(ctmp2,9002) ' we can eliminate only ', jpni*jpnj - inijmin, ' land mpi subdomains therefore ' |
|---|
| 265 | WRITE(ctmp3,9001) ' the number of ocean mpi subdomains (', inijmin,') exceed the number of MPI processes:', mppsize |
|---|
| 266 | WRITE(ctmp4,*) ' ==>>> There is the list of best domain decompositions you should use: ' |
|---|
| 267 | CALL ctl_stop( ctmp1, ctmp2, ctmp3, ' ', ctmp4, ' ' ) |
|---|
| 268 | CALL bestpartition( mppsize, ldlist = .TRUE. ) ! must be done by all core |
|---|
| 269 | ENDIF |
|---|
| 270 | |
|---|
| 271 | IF( mppsize > jpni*jpnj ) THEN ! not enough mpi subdomains for the total number of mpi processes |
|---|
| 272 | IF(lwp) THEN |
|---|
| 273 | WRITE(numout,9003) ' The number of mpi processes: ', mppsize |
|---|
| 274 | WRITE(numout,9003) ' exceeds the maximum number of subdomains (ocean+land) = ', jpni*jpnj |
|---|
| 275 | WRITE(numout,9001) ' defined by the following domain decomposition: jpni = ', jpni, ' jpnj = ', jpnj |
|---|
| 276 | WRITE(numout, *) ' You should: ' |
|---|
| 277 | IF( llauto ) THEN |
|---|
| 278 | WRITE(numout,*) ' - either prescribe your domain decomposition with the namelist variables' |
|---|
| 279 | WRITE(numout,*) ' jpni and jpnj to match the number of mpi process you want to use, ' |
|---|
| 280 | WRITE(numout,*) ' even IF it not the best choice...' |
|---|
| 281 | WRITE(numout,*) ' - or keep the automatic and optimal domain decomposition by picking up one' |
|---|
| 282 | WRITE(numout,*) ' of the number of mpi process proposed in the list bellow' |
|---|
| 283 | ELSE |
|---|
| 284 | WRITE(numout,*) ' - either properly prescribe your domain decomposition with jpni and jpnj' |
|---|
| 285 | WRITE(numout,*) ' in order to be consistent with the number of mpi process you want to use' |
|---|
| 286 | WRITE(numout,*) ' even IF it not the best choice...' |
|---|
| 287 | WRITE(numout,*) ' - or use the automatic and optimal domain decomposition and pick up one of' |
|---|
| 288 | WRITE(numout,*) ' the domain decomposition proposed in the list bellow' |
|---|
| 289 | ENDIF |
|---|
| 290 | WRITE(numout,*) |
|---|
| 291 | ENDIF |
|---|
| 292 | CALL bestpartition( mppsize, ldlist = .TRUE. ) ! must be done by all core |
|---|
| 293 | ENDIF |
|---|
| 294 | |
|---|
| 295 | jpnij = mppsize ! force jpnij definition <-- remove as much land subdomains as needed to reach this condition |
|---|
| 296 | IF( mppsize > inijmin ) THEN |
|---|
| 297 | WRITE(ctmp1,9003) ' The number of mpi processes: ', mppsize |
|---|
| 298 | WRITE(ctmp2,9003) ' exceeds the maximum number of ocean subdomains = ', inijmin |
|---|
| 299 | WRITE(ctmp3,9002) ' we suppressed ', jpni*jpnj - mppsize, ' land subdomains ' |
|---|
| 300 | WRITE(ctmp4,9002) ' BUT we had to keep ', mppsize - inijmin, ' land subdomains that are useless...' |
|---|
| 301 | CALL ctl_warn( ctmp1, ctmp2, ctmp3, ctmp4, ' ', ' --- YOU ARE WASTING CPU... ---', ' ' ) |
|---|
| 302 | ELSE ! mppsize = inijmin |
|---|
| 303 | IF(lwp) THEN |
|---|
| 304 | IF(llbest) WRITE(numout,*) ' ==> you use the best mpi decomposition' |
|---|
| 305 | WRITE(numout,*) |
|---|
| 306 | WRITE(numout,9003) ' Number of mpi processes: ', mppsize |
|---|
| 307 | WRITE(numout,9003) ' Number of ocean subdomains = ', inijmin |
|---|
| 308 | WRITE(numout,9003) ' Number of suppressed land subdomains = ', jpni*jpnj - inijmin |
|---|
| 309 | WRITE(numout,*) |
|---|
| 310 | ENDIF |
|---|
| 311 | ENDIF |
|---|
| 312 | 9000 FORMAT (a, i4, a, i4, a, i7, a) |
|---|
| 313 | 9001 FORMAT (a, i4, a, i4) |
|---|
| 314 | 9002 FORMAT (a, i4, a) |
|---|
| 315 | 9003 FORMAT (a, i5) |
|---|
| 316 | |
|---|
| 317 | IF( numbot /= -1 ) CALL iom_close( numbot ) |
|---|
| 318 | IF( numbdy /= -1 ) CALL iom_close( numbdy ) |
|---|
| 319 | |
|---|
| 320 | ALLOCATE( nfiimpp(jpni,jpnj), nfipproc(jpni,jpnj), nfijpit(jpni,jpnj) , & |
|---|
| 321 | & nimppt(jpnij) , ibonit(jpnij) , jpiall(jpnij) , jpjall(jpnij) , & |
|---|
| 322 | & njmppt(jpnij) , ibonjt(jpnij) , nis0all(jpnij) , njs0all(jpnij) , & |
|---|
| 323 | & nie0all(jpnij) , nje0all(jpnij) , & |
|---|
| 324 | & iin(jpnij), ii_nono(jpnij), ii_noea(jpnij), & |
|---|
| 325 | & ijn(jpnij), ii_noso(jpnij), ii_nowe(jpnij), & |
|---|
| 326 | & iimppt(jpni,jpnj), ijpi(jpni,jpnj), ibondi(jpni,jpnj), ipproc(jpni,jpnj), & |
|---|
| 327 | & ijmppt(jpni,jpnj), ijpj(jpni,jpnj), ibondj(jpni,jpnj), ipolj(jpni,jpnj), & |
|---|
| 328 | & iie0(jpni,jpnj), iis0(jpni,jpnj), iono(jpni,jpnj), ioea(jpni,jpnj), & |
|---|
| 329 | & ije0(jpni,jpnj), ijs0(jpni,jpnj), ioso(jpni,jpnj), iowe(jpni,jpnj), & |
|---|
| 330 | & STAT=ierr ) |
|---|
| 331 | CALL mpp_sum( 'mppini', ierr ) |
|---|
| 332 | IF( ierr /= 0 ) CALL ctl_stop( 'STOP', 'mpp_init: unable to allocate standard ocean arrays' ) |
|---|
| 333 | |
|---|
| 334 | #if defined key_agrif |
|---|
| 335 | IF( .NOT. Agrif_Root() ) THEN ! AGRIF children: specific setting (cf. agrif_user.F90) |
|---|
| 336 | IF( jpiglo /= nbcellsx + 2 + 2*nbghostcells ) & |
|---|
| 337 | CALL ctl_stop( 'STOP', 'mpp_init: Agrif children requires jpiglo == nbcellsx + 2 + 2*nbghostcells' ) |
|---|
| 338 | IF( jpjglo /= nbcellsy + 2 + 2*nbghostcells ) & |
|---|
| 339 | CALL ctl_stop( 'STOP', 'mpp_init: Agrif children requires jpjglo == nbcellsy + 2 + 2*nbghostcells' ) |
|---|
| 340 | IF( ln_use_jattr ) CALL ctl_stop( 'STOP', 'mpp_init:Agrif children requires ln_use_jattr = .false. ' ) |
|---|
| 341 | ENDIF |
|---|
| 342 | #endif |
|---|
| 343 | ! |
|---|
| 344 | ! 2. Index arrays for subdomains |
|---|
| 345 | ! ----------------------------------- |
|---|
| 346 | ! |
|---|
| 347 | CALL basic_decomposition( jpiglo, jpjglo, nn_hls, jpni, jpnj, jpimax, jpjmax, iimppt, ijmppt, ijpi, ijpj ) |
|---|
| 348 | nfiimpp(:,:) = iimppt(:,:) |
|---|
| 349 | nfijpit(:,:) = ijpi(:,:) |
|---|
| 350 | ! |
|---|
| 351 | IF(lwp) THEN |
|---|
| 352 | WRITE(numout,*) |
|---|
| 353 | WRITE(numout,*) 'MPI Message Passing MPI - domain lay out over processors' |
|---|
| 354 | WRITE(numout,*) |
|---|
| 355 | WRITE(numout,*) ' defines mpp subdomains' |
|---|
| 356 | WRITE(numout,*) ' jpni = ', jpni |
|---|
| 357 | WRITE(numout,*) ' jpnj = ', jpnj |
|---|
| 358 | WRITE(numout,*) |
|---|
| 359 | WRITE(numout,*) ' sum ijpi(i,1) = ', sum(ijpi(:,1)), ' jpiglo = ', jpiglo |
|---|
| 360 | WRITE(numout,*) ' sum ijpj(1,j) = ', sum(ijpj(1,:)), ' jpjglo = ', jpjglo |
|---|
| 361 | ENDIF |
|---|
| 362 | |
|---|
| 363 | ! 3. Subdomain description in the Regular Case |
|---|
| 364 | ! -------------------------------------------- |
|---|
| 365 | ! specific cases where there is no communication -> must do the periodicity by itself |
|---|
| 366 | ! Warning: because of potential land-area suppression, do not use nbond[ij] == 2 |
|---|
| 367 | l_Iperio = jpni == 1 .AND. (jperio == 1 .OR. jperio == 4 .OR. jperio == 6 .OR. jperio == 7) |
|---|
| 368 | l_Jperio = jpnj == 1 .AND. (jperio == 2 .OR. jperio == 7) |
|---|
| 369 | |
|---|
| 370 | DO jarea = 1, jpni*jpnj |
|---|
| 371 | ! |
|---|
| 372 | iarea0 = jarea - 1 |
|---|
| 373 | ii = 1 + MOD(iarea0,jpni) |
|---|
| 374 | ij = 1 + iarea0/jpni |
|---|
| 375 | ili = ijpi(ii,ij) |
|---|
| 376 | ilj = ijpj(ii,ij) |
|---|
| 377 | ibondi(ii,ij) = 0 ! default: has e-w neighbours |
|---|
| 378 | IF( ii == 1 ) ibondi(ii,ij) = -1 ! first column, has only e neighbour |
|---|
| 379 | IF( ii == jpni ) ibondi(ii,ij) = 1 ! last column, has only w neighbour |
|---|
| 380 | IF( jpni == 1 ) ibondi(ii,ij) = 2 ! has no e-w neighbour |
|---|
| 381 | ibondj(ii,ij) = 0 ! default: has n-s neighbours |
|---|
| 382 | IF( ij == 1 ) ibondj(ii,ij) = -1 ! first row, has only n neighbour |
|---|
| 383 | IF( ij == jpnj ) ibondj(ii,ij) = 1 ! last row, has only s neighbour |
|---|
| 384 | IF( jpnj == 1 ) ibondj(ii,ij) = 2 ! has no n-s neighbour |
|---|
| 385 | |
|---|
| 386 | ! Subdomain neighbors (get their zone number): default definition |
|---|
| 387 | ioso(ii,ij) = iarea0 - jpni |
|---|
| 388 | iowe(ii,ij) = iarea0 - 1 |
|---|
| 389 | ioea(ii,ij) = iarea0 + 1 |
|---|
| 390 | iono(ii,ij) = iarea0 + jpni |
|---|
| 391 | iis0(ii,ij) = 1 + nn_hls |
|---|
| 392 | iie0(ii,ij) = ili - nn_hls |
|---|
| 393 | ijs0(ii,ij) = 1 + nn_hls |
|---|
| 394 | ije0(ii,ij) = ilj - nn_hls |
|---|
| 395 | |
|---|
| 396 | ! East-West periodicity: change ibondi, ioea, iowe |
|---|
| 397 | IF( jperio == 1 .OR. jperio == 4 .OR. jperio == 6 .OR. jperio == 7 ) THEN |
|---|
| 398 | IF( jpni /= 1 ) ibondi(ii,ij) = 0 ! redefine: all have e-w neighbours |
|---|
| 399 | IF( ii == 1 ) iowe(ii,ij) = iarea0 + (jpni-1) ! redefine: first column, address of w neighbour |
|---|
| 400 | IF( ii == jpni ) ioea(ii,ij) = iarea0 - (jpni-1) ! redefine: last column, address of e neighbour |
|---|
| 401 | ENDIF |
|---|
| 402 | |
|---|
| 403 | ! Simple North-South periodicity: change ibondj, ioso, iono |
|---|
| 404 | IF( jperio == 2 .OR. jperio == 7 ) THEN |
|---|
| 405 | IF( jpnj /= 1 ) ibondj(ii,ij) = 0 ! redefine: all have n-s neighbours |
|---|
| 406 | IF( ij == 1 ) ioso(ii,ij) = iarea0 + jpni * (jpnj-1) ! redefine: first row, address of s neighbour |
|---|
| 407 | IF( ij == jpnj ) iono(ii,ij) = iarea0 - jpni * (jpnj-1) ! redefine: last row, address of n neighbour |
|---|
| 408 | ENDIF |
|---|
| 409 | |
|---|
| 410 | ! North fold: define ipolj, change iono. Warning: we do not change ibondj... |
|---|
| 411 | ipolj(ii,ij) = 0 |
|---|
| 412 | IF( jperio == 3 .OR. jperio == 4 ) THEN |
|---|
| 413 | ijm1 = jpni*(jpnj-1) |
|---|
| 414 | imil = ijm1+(jpni+1)/2 |
|---|
| 415 | IF( jarea > ijm1 ) ipolj(ii,ij) = 3 |
|---|
| 416 | IF( MOD(jpni,2) == 1 .AND. jarea == imil ) ipolj(ii,ij) = 4 |
|---|
| 417 | IF( ipolj(ii,ij) == 3 ) iono(ii,ij) = jpni*jpnj-jarea+ijm1 ! MPI rank of northern neighbour |
|---|
| 418 | ENDIF |
|---|
| 419 | IF( jperio == 5 .OR. jperio == 6 ) THEN |
|---|
| 420 | ijm1 = jpni*(jpnj-1) |
|---|
| 421 | imil = ijm1+(jpni+1)/2 |
|---|
| 422 | IF( jarea > ijm1) ipolj(ii,ij) = 5 |
|---|
| 423 | IF( MOD(jpni,2) == 1 .AND. jarea == imil ) ipolj(ii,ij) = 6 |
|---|
| 424 | IF( ipolj(ii,ij) == 5) iono(ii,ij) = jpni*jpnj-jarea+ijm1 ! MPI rank of northern neighbour |
|---|
| 425 | ENDIF |
|---|
| 426 | ! |
|---|
| 427 | END DO |
|---|
| 428 | |
|---|
| 429 | ! 4. deal with land subdomains |
|---|
| 430 | ! ---------------------------- |
|---|
| 431 | ! |
|---|
| 432 | ! specify which subdomains are oce subdomains; other are land subdomains |
|---|
| 433 | ipproc(:,:) = -1 |
|---|
| 434 | icont = -1 |
|---|
| 435 | DO jarea = 1, jpni*jpnj |
|---|
| 436 | iarea0 = jarea - 1 |
|---|
| 437 | ii = 1 + MOD(iarea0,jpni) |
|---|
| 438 | ij = 1 + iarea0/jpni |
|---|
| 439 | IF( llisoce(ii,ij) ) THEN |
|---|
| 440 | icont = icont + 1 |
|---|
| 441 | ipproc(ii,ij) = icont |
|---|
| 442 | iin(icont+1) = ii |
|---|
| 443 | ijn(icont+1) = ij |
|---|
| 444 | ENDIF |
|---|
| 445 | END DO |
|---|
| 446 | ! if needed add some land subdomains to reach jpnij active subdomains |
|---|
| 447 | i2add = jpnij - inijmin |
|---|
| 448 | DO jarea = 1, jpni*jpnj |
|---|
| 449 | iarea0 = jarea - 1 |
|---|
| 450 | ii = 1 + MOD(iarea0,jpni) |
|---|
| 451 | ij = 1 + iarea0/jpni |
|---|
| 452 | IF( .NOT. llisoce(ii,ij) .AND. i2add > 0 ) THEN |
|---|
| 453 | icont = icont + 1 |
|---|
| 454 | ipproc(ii,ij) = icont |
|---|
| 455 | iin(icont+1) = ii |
|---|
| 456 | ijn(icont+1) = ij |
|---|
| 457 | i2add = i2add - 1 |
|---|
| 458 | ENDIF |
|---|
| 459 | END DO |
|---|
| 460 | nfipproc(:,:) = ipproc(:,:) |
|---|
| 461 | |
|---|
| 462 | ! neighbour treatment: change ibondi, ibondj if next to a land zone |
|---|
| 463 | DO jarea = 1, jpni*jpnj |
|---|
| 464 | ii = 1 + MOD( jarea-1 , jpni ) |
|---|
| 465 | ij = 1 + (jarea-1) / jpni |
|---|
| 466 | ! land-only area with an active n neigbour |
|---|
| 467 | IF ( ipproc(ii,ij) == -1 .AND. 0 <= iono(ii,ij) .AND. iono(ii,ij) <= jpni*jpnj-1 ) THEN |
|---|
| 468 | iino = 1 + MOD( iono(ii,ij) , jpni ) ! ii index of this n neigbour |
|---|
| 469 | ijno = 1 + iono(ii,ij) / jpni ! ij index of this n neigbour |
|---|
| 470 | ! In case of north fold exchange: I am the n neigbour of my n neigbour!! (#1057) |
|---|
| 471 | ! --> for northern neighbours of northern row processors (in case of north-fold) |
|---|
| 472 | ! need to reverse the LOGICAL direction of communication |
|---|
| 473 | idir = 1 ! we are indeed the s neigbour of this n neigbour |
|---|
| 474 | IF( ij == jpnj .AND. ijno == jpnj ) idir = -1 ! both are on the last row, we are in fact the n neigbour |
|---|
| 475 | IF( ibondj(iino,ijno) == idir ) ibondj(iino,ijno) = 2 ! this n neigbour had only a s/n neigbour -> no more |
|---|
| 476 | IF( ibondj(iino,ijno) == 0 ) ibondj(iino,ijno) = -idir ! this n neigbour had both, n-s neighbours -> keep 1 |
|---|
| 477 | ENDIF |
|---|
| 478 | ! land-only area with an active s neigbour |
|---|
| 479 | IF( ipproc(ii,ij) == -1 .AND. 0 <= ioso(ii,ij) .AND. ioso(ii,ij) <= jpni*jpnj-1 ) THEN |
|---|
| 480 | iiso = 1 + MOD( ioso(ii,ij) , jpni ) ! ii index of this s neigbour |
|---|
| 481 | ijso = 1 + ioso(ii,ij) / jpni ! ij index of this s neigbour |
|---|
| 482 | IF( ibondj(iiso,ijso) == -1 ) ibondj(iiso,ijso) = 2 ! this s neigbour had only a n neigbour -> no more neigbour |
|---|
| 483 | IF( ibondj(iiso,ijso) == 0 ) ibondj(iiso,ijso) = 1 ! this s neigbour had both, n-s neighbours -> keep s neigbour |
|---|
| 484 | ENDIF |
|---|
| 485 | ! land-only area with an active e neigbour |
|---|
| 486 | IF( ipproc(ii,ij) == -1 .AND. 0 <= ioea(ii,ij) .AND. ioea(ii,ij) <= jpni*jpnj-1 ) THEN |
|---|
| 487 | iiea = 1 + MOD( ioea(ii,ij) , jpni ) ! ii index of this e neigbour |
|---|
| 488 | ijea = 1 + ioea(ii,ij) / jpni ! ij index of this e neigbour |
|---|
| 489 | IF( ibondi(iiea,ijea) == 1 ) ibondi(iiea,ijea) = 2 ! this e neigbour had only a w neigbour -> no more neigbour |
|---|
| 490 | IF( ibondi(iiea,ijea) == 0 ) ibondi(iiea,ijea) = -1 ! this e neigbour had both, e-w neighbours -> keep e neigbour |
|---|
| 491 | ENDIF |
|---|
| 492 | ! land-only area with an active w neigbour |
|---|
| 493 | IF( ipproc(ii,ij) == -1 .AND. 0 <= iowe(ii,ij) .AND. iowe(ii,ij) <= jpni*jpnj-1) THEN |
|---|
| 494 | iiwe = 1 + MOD( iowe(ii,ij) , jpni ) ! ii index of this w neigbour |
|---|
| 495 | ijwe = 1 + iowe(ii,ij) / jpni ! ij index of this w neigbour |
|---|
| 496 | IF( ibondi(iiwe,ijwe) == -1 ) ibondi(iiwe,ijwe) = 2 ! this w neigbour had only a e neigbour -> no more neigbour |
|---|
| 497 | IF( ibondi(iiwe,ijwe) == 0 ) ibondi(iiwe,ijwe) = 1 ! this w neigbour had both, e-w neighbours -> keep w neigbour |
|---|
| 498 | ENDIF |
|---|
| 499 | END DO |
|---|
| 500 | |
|---|
| 501 | ! 5. Subdomain print |
|---|
| 502 | ! ------------------ |
|---|
| 503 | IF(lwp) THEN |
|---|
| 504 | ifreq = 4 |
|---|
| 505 | il1 = 1 |
|---|
| 506 | DO jn = 1, (jpni-1)/ifreq+1 |
|---|
| 507 | il2 = MIN(jpni,il1+ifreq-1) |
|---|
| 508 | WRITE(numout,*) |
|---|
| 509 | WRITE(numout,9400) ('***',ji=il1,il2-1) |
|---|
| 510 | DO jj = jpnj, 1, -1 |
|---|
| 511 | WRITE(numout,9403) (' ',ji=il1,il2-1) |
|---|
| 512 | WRITE(numout,9402) jj, (ijpi(ji,jj),ijpj(ji,jj),ji=il1,il2) |
|---|
| 513 | WRITE(numout,9404) (ipproc(ji,jj),ji=il1,il2) |
|---|
| 514 | WRITE(numout,9403) (' ',ji=il1,il2-1) |
|---|
| 515 | WRITE(numout,9400) ('***',ji=il1,il2-1) |
|---|
| 516 | END DO |
|---|
| 517 | WRITE(numout,9401) (ji,ji=il1,il2) |
|---|
| 518 | il1 = il1+ifreq |
|---|
| 519 | END DO |
|---|
| 520 | 9400 FORMAT(' ***' ,20('*************',a3) ) |
|---|
| 521 | 9403 FORMAT(' * ',20(' * ',a3) ) |
|---|
| 522 | 9401 FORMAT(' ' ,20(' ',i3,' ') ) |
|---|
| 523 | 9402 FORMAT(' ',i3,' * ',20(i3,' x',i3,' * ') ) |
|---|
| 524 | 9404 FORMAT(' * ' ,20(' ' ,i4,' * ') ) |
|---|
| 525 | ENDIF |
|---|
| 526 | |
|---|
| 527 | ! just to save nono etc for all proc |
|---|
| 528 | ! warning ii*ij (zone) /= nproc (processors)! |
|---|
| 529 | ! ioso = zone number, ii_noso = proc number |
|---|
| 530 | ii_noso(:) = -1 |
|---|
| 531 | ii_nono(:) = -1 |
|---|
| 532 | ii_noea(:) = -1 |
|---|
| 533 | ii_nowe(:) = -1 |
|---|
| 534 | DO jproc = 1, jpnij |
|---|
| 535 | ii = iin(jproc) |
|---|
| 536 | ij = ijn(jproc) |
|---|
| 537 | IF( 0 <= ioso(ii,ij) .AND. ioso(ii,ij) <= (jpni*jpnj-1) ) THEN |
|---|
| 538 | iiso = 1 + MOD( ioso(ii,ij) , jpni ) |
|---|
| 539 | ijso = 1 + ioso(ii,ij) / jpni |
|---|
| 540 | ii_noso(jproc) = ipproc(iiso,ijso) |
|---|
| 541 | ENDIF |
|---|
| 542 | IF( 0 <= iowe(ii,ij) .AND. iowe(ii,ij) <= (jpni*jpnj-1) ) THEN |
|---|
| 543 | iiwe = 1 + MOD( iowe(ii,ij) , jpni ) |
|---|
| 544 | ijwe = 1 + iowe(ii,ij) / jpni |
|---|
| 545 | ii_nowe(jproc) = ipproc(iiwe,ijwe) |
|---|
| 546 | ENDIF |
|---|
| 547 | IF( 0 <= ioea(ii,ij) .AND. ioea(ii,ij) <= (jpni*jpnj-1) ) THEN |
|---|
| 548 | iiea = 1 + MOD( ioea(ii,ij) , jpni ) |
|---|
| 549 | ijea = 1 + ioea(ii,ij) / jpni |
|---|
| 550 | ii_noea(jproc)= ipproc(iiea,ijea) |
|---|
| 551 | ENDIF |
|---|
| 552 | IF( 0 <= iono(ii,ij) .AND. iono(ii,ij) <= (jpni*jpnj-1) ) THEN |
|---|
| 553 | iino = 1 + MOD( iono(ii,ij) , jpni ) |
|---|
| 554 | ijno = 1 + iono(ii,ij) / jpni |
|---|
| 555 | ii_nono(jproc)= ipproc(iino,ijno) |
|---|
| 556 | ENDIF |
|---|
| 557 | END DO |
|---|
| 558 | |
|---|
| 559 | ! 6. Change processor name |
|---|
| 560 | ! ------------------------ |
|---|
| 561 | ii = iin(narea) |
|---|
| 562 | ij = ijn(narea) |
|---|
| 563 | ! |
|---|
| 564 | ! set default neighbours |
|---|
| 565 | noso = ii_noso(narea) |
|---|
| 566 | nowe = ii_nowe(narea) |
|---|
| 567 | noea = ii_noea(narea) |
|---|
| 568 | nono = ii_nono(narea) |
|---|
| 569 | jpi = ijpi(ii,ij) |
|---|
| 570 | !!$ Nis0 = iis0(ii,ij) |
|---|
| 571 | !!$ Nie0 = iie0(ii,ij) |
|---|
| 572 | jpj = ijpj(ii,ij) |
|---|
| 573 | !!$ Njs0 = ijs0(ii,ij) |
|---|
| 574 | !!$ Nje0 = ije0(ii,ij) |
|---|
| 575 | nbondi = ibondi(ii,ij) |
|---|
| 576 | nbondj = ibondj(ii,ij) |
|---|
| 577 | nimpp = iimppt(ii,ij) |
|---|
| 578 | njmpp = ijmppt(ii,ij) |
|---|
| 579 | jpk = jpkglo ! third dim |
|---|
| 580 | #if defined key_agrif |
|---|
| 581 | ! simple trick to use same vertical grid as parent but different number of levels: |
|---|
| 582 | ! Save maximum number of levels in jpkglo, then define all vertical grids with this number. |
|---|
| 583 | ! Suppress once vertical online interpolation is ok |
|---|
| 584 | !!$ IF(.NOT.Agrif_Root()) jpkglo = Agrif_Parent( jpkglo ) |
|---|
| 585 | #endif |
|---|
| 586 | ! |
|---|
| 587 | CALL init_doloop ! set start/end indices of do-loop, depending on the halo width value (nn_hls) |
|---|
| 588 | ! |
|---|
| 589 | jpim1 = jpi-1 ! inner domain indices |
|---|
| 590 | jpjm1 = jpj-1 ! " " |
|---|
| 591 | jpkm1 = MAX( 1, jpk-1 ) ! " " |
|---|
| 592 | jpij = jpi*jpj ! jpi x j |
|---|
| 593 | DO jproc = 1, jpnij |
|---|
| 594 | ii = iin(jproc) |
|---|
| 595 | ij = ijn(jproc) |
|---|
| 596 | jpiall (jproc) = ijpi(ii,ij) |
|---|
| 597 | nis0all(jproc) = iis0(ii,ij) |
|---|
| 598 | nie0all(jproc) = iie0(ii,ij) |
|---|
| 599 | jpjall (jproc) = ijpj(ii,ij) |
|---|
| 600 | njs0all(jproc) = ijs0(ii,ij) |
|---|
| 601 | nje0all(jproc) = ije0(ii,ij) |
|---|
| 602 | ibonit(jproc) = ibondi(ii,ij) |
|---|
| 603 | ibonjt(jproc) = ibondj(ii,ij) |
|---|
| 604 | nimppt(jproc) = iimppt(ii,ij) |
|---|
| 605 | njmppt(jproc) = ijmppt(ii,ij) |
|---|
| 606 | END DO |
|---|
| 607 | |
|---|
| 608 | ! Save processor layout in ascii file |
|---|
| 609 | IF (llwrtlay) THEN |
|---|
| 610 | CALL ctl_opn( inum, 'layout.dat', 'REPLACE', 'FORMATTED', 'SEQUENTIAL', -1, numout, .FALSE., narea ) |
|---|
| 611 | WRITE(inum,'(a)') ' jpnij jpimax jpjmax jpk jpiglo jpjglo'//& |
|---|
| 612 | & ' ( local: narea jpi jpj )' |
|---|
| 613 | WRITE(inum,'(6i8,a,3i8,a)') jpnij,jpimax,jpjmax,jpk,jpiglo,jpjglo,& |
|---|
| 614 | & ' ( local: ',narea,jpi,jpj,' )' |
|---|
| 615 | WRITE(inum,'(a)') 'nproc jpi jpj Nis0 Njs0 Nie0 Nje0 nimp njmp nono noso nowe noea nbondi nbondj ' |
|---|
| 616 | |
|---|
| 617 | DO jproc = 1, jpnij |
|---|
| 618 | WRITE(inum,'(13i5,2i7)') jproc-1, jpiall(jproc), jpjall(jproc), & |
|---|
| 619 | & nis0all(jproc), njs0all(jproc), & |
|---|
| 620 | & nie0all(jproc), nje0all(jproc), & |
|---|
| 621 | & nimppt (jproc), njmppt (jproc), & |
|---|
| 622 | & ii_nono(jproc), ii_noso(jproc), & |
|---|
| 623 | & ii_nowe(jproc), ii_noea(jproc), & |
|---|
| 624 | & ibonit (jproc), ibonjt (jproc) |
|---|
| 625 | END DO |
|---|
| 626 | END IF |
|---|
| 627 | |
|---|
| 628 | ! ! north fold parameter |
|---|
| 629 | ! Defined npolj, either 0, 3 , 4 , 5 , 6 |
|---|
| 630 | ! In this case the important thing is that npolj /= 0 |
|---|
| 631 | ! Because if we go through these line it is because jpni >1 and thus |
|---|
| 632 | ! we must use lbcnorthmpp, which tests only npolj =0 or npolj /= 0 |
|---|
| 633 | npolj = 0 |
|---|
| 634 | ij = ijn(narea) |
|---|
| 635 | IF( jperio == 3 .OR. jperio == 4 ) THEN |
|---|
| 636 | IF( ij == jpnj ) npolj = 3 |
|---|
| 637 | ENDIF |
|---|
| 638 | IF( jperio == 5 .OR. jperio == 6 ) THEN |
|---|
| 639 | IF( ij == jpnj ) npolj = 5 |
|---|
| 640 | ENDIF |
|---|
| 641 | ! |
|---|
| 642 | nproc = narea-1 |
|---|
| 643 | IF(lwp) THEN |
|---|
| 644 | WRITE(numout,*) |
|---|
| 645 | WRITE(numout,*) ' resulting internal parameters : ' |
|---|
| 646 | WRITE(numout,*) ' nproc = ', nproc |
|---|
| 647 | WRITE(numout,*) ' nowe = ', nowe , ' noea = ', noea |
|---|
| 648 | WRITE(numout,*) ' nono = ', nono , ' noso = ', noso |
|---|
| 649 | WRITE(numout,*) ' nbondi = ', nbondi |
|---|
| 650 | WRITE(numout,*) ' nbondj = ', nbondj |
|---|
| 651 | WRITE(numout,*) ' npolj = ', npolj |
|---|
| 652 | WRITE(numout,*) ' l_Iperio = ', l_Iperio |
|---|
| 653 | WRITE(numout,*) ' l_Jperio = ', l_Jperio |
|---|
| 654 | WRITE(numout,*) ' nimpp = ', nimpp |
|---|
| 655 | WRITE(numout,*) ' njmpp = ', njmpp |
|---|
| 656 | WRITE(numout,*) ' nn_hls = ', nn_hls |
|---|
| 657 | ENDIF |
|---|
| 658 | |
|---|
| 659 | ! ! Prepare mpp north fold |
|---|
| 660 | IF( jperio >= 3 .AND. jperio <= 6 .AND. jpni > 1 ) THEN |
|---|
| 661 | CALL mpp_ini_north |
|---|
| 662 | IF (lwp) THEN |
|---|
| 663 | WRITE(numout,*) |
|---|
| 664 | WRITE(numout,*) ' ==>>> North fold boundary prepared for jpni >1' |
|---|
| 665 | ! additional prints in layout.dat |
|---|
| 666 | ENDIF |
|---|
| 667 | IF (llwrtlay) THEN |
|---|
| 668 | WRITE(inum,*) |
|---|
| 669 | WRITE(inum,*) |
|---|
| 670 | WRITE(inum,*) 'number of subdomains located along the north fold : ', ndim_rank_north |
|---|
| 671 | WRITE(inum,*) 'Rank of the subdomains located along the north fold : ', ndim_rank_north |
|---|
| 672 | DO jproc = 1, ndim_rank_north, 5 |
|---|
| 673 | WRITE(inum,*) nrank_north( jproc:MINVAL( (/jproc+4,ndim_rank_north/) ) ) |
|---|
| 674 | END DO |
|---|
| 675 | ENDIF |
|---|
| 676 | ENDIF |
|---|
| 677 | ! |
|---|
| 678 | CALL init_ioipsl ! Prepare NetCDF output file (if necessary) |
|---|
| 679 | ! |
|---|
| 680 | IF (( jperio >= 3 .AND. jperio <= 6 .AND. jpni > 1 ).AND.( ln_nnogather )) THEN |
|---|
| 681 | CALL init_nfdcom ! northfold neighbour lists |
|---|
| 682 | IF (llwrtlay) THEN |
|---|
| 683 | WRITE(inum,*) |
|---|
| 684 | WRITE(inum,*) |
|---|
| 685 | WRITE(inum,*) 'north fold exchanges with explicit point-to-point messaging :' |
|---|
| 686 | WRITE(inum,*) 'nsndto : ', nsndto |
|---|
| 687 | WRITE(inum,*) 'isendto : ', isendto |
|---|
| 688 | ENDIF |
|---|
| 689 | ENDIF |
|---|
| 690 | ! |
|---|
| 691 | IF (llwrtlay) CLOSE(inum) |
|---|
| 692 | ! |
|---|
| 693 | DEALLOCATE(iin, ijn, ii_nono, ii_noea, ii_noso, ii_nowe, & |
|---|
| 694 | & iimppt, ijmppt, ibondi, ibondj, ipproc, ipolj, & |
|---|
| 695 | & ijpi, ijpj, iie0, ije0, iis0, ijs0, & |
|---|
| 696 | & iono, ioea, ioso, iowe, llisoce) |
|---|
| 697 | ! |
|---|
| 698 | END SUBROUTINE mpp_init |
|---|
| 699 | |
|---|
| 700 | |
|---|
| 701 | SUBROUTINE basic_decomposition( kiglo, kjglo, khls, knbi, knbj, kimax, kjmax, kimppt, kjmppt, klci, klcj) |
|---|
| 702 | !!---------------------------------------------------------------------- |
|---|
| 703 | !! *** ROUTINE basic_decomposition *** |
|---|
| 704 | !! |
|---|
| 705 | !! ** Purpose : Lay out the global domain over processors. |
|---|
| 706 | !! |
|---|
| 707 | !! ** Method : Global domain is distributed in smaller local domains. |
|---|
| 708 | !! |
|---|
| 709 | !! ** Action : - set for all knbi*knbj domains: |
|---|
| 710 | !! kimppt : longitudinal index |
|---|
| 711 | !! kjmppt : latitudinal index |
|---|
| 712 | !! klci : first dimension |
|---|
| 713 | !! klcj : second dimension |
|---|
| 714 | !!---------------------------------------------------------------------- |
|---|
| 715 | INTEGER, INTENT(in ) :: kiglo, kjglo |
|---|
| 716 | INTEGER, INTENT(in ) :: khls |
|---|
| 717 | INTEGER, INTENT(in ) :: knbi, knbj |
|---|
| 718 | INTEGER, INTENT( out) :: kimax, kjmax |
|---|
| 719 | INTEGER, DIMENSION(knbi,knbj), OPTIONAL, INTENT( out) :: kimppt, kjmppt |
|---|
| 720 | INTEGER, DIMENSION(knbi,knbj), OPTIONAL, INTENT( out) :: klci, klcj |
|---|
| 721 | ! |
|---|
| 722 | INTEGER :: ji, jj |
|---|
| 723 | INTEGER :: i2hls |
|---|
| 724 | INTEGER :: iresti, irestj, irm, ijpjmin |
|---|
| 725 | !!---------------------------------------------------------------------- |
|---|
| 726 | i2hls = 2*khls |
|---|
| 727 | ! |
|---|
| 728 | #if defined key_nemocice_decomp |
|---|
| 729 | kimax = ( nx_global+2-i2hls + (knbi-1) ) / knbi + i2hls ! first dim. |
|---|
| 730 | kjmax = ( ny_global+2-i2hls + (knbj-1) ) / knbj + i2hls ! second dim. |
|---|
| 731 | #else |
|---|
| 732 | kimax = ( kiglo - i2hls + (knbi-1) ) / knbi + i2hls ! first dim. |
|---|
| 733 | kjmax = ( kjglo - i2hls + (knbj-1) ) / knbj + i2hls ! second dim. |
|---|
| 734 | #endif |
|---|
| 735 | IF( .NOT. PRESENT(kimppt) ) RETURN |
|---|
| 736 | ! |
|---|
| 737 | ! 1. Dimension arrays for subdomains |
|---|
| 738 | ! ----------------------------------- |
|---|
| 739 | ! Computation of local domain sizes klci() klcj() |
|---|
| 740 | ! These dimensions depend on global sizes knbi,knbj and kiglo,kjglo |
|---|
| 741 | ! The subdomains are squares lesser than or equal to the global |
|---|
| 742 | ! dimensions divided by the number of processors minus the overlap array. |
|---|
| 743 | ! |
|---|
| 744 | iresti = 1 + MOD( kiglo - i2hls - 1 , knbi ) |
|---|
| 745 | irestj = 1 + MOD( kjglo - i2hls - 1 , knbj ) |
|---|
| 746 | ! |
|---|
| 747 | ! Need to use kimax and kjmax here since jpi and jpj not yet defined |
|---|
| 748 | #if defined key_nemocice_decomp |
|---|
| 749 | ! Change padding to be consistent with CICE |
|---|
| 750 | klci(1:knbi-1,: ) = kimax |
|---|
| 751 | klci( knbi ,: ) = kiglo - (knbi - 1) * (kimax - i2hls) |
|---|
| 752 | klcj(: ,1:knbj-1) = kjmax |
|---|
| 753 | klcj(: , knbj ) = kjglo - (knbj - 1) * (kjmax - i2hls) |
|---|
| 754 | #else |
|---|
| 755 | klci(1:iresti ,:) = kimax |
|---|
| 756 | klci(iresti+1:knbi ,:) = kimax-1 |
|---|
| 757 | IF( MINVAL(klci) < 2*i2hls ) THEN |
|---|
| 758 | WRITE(ctmp1,*) ' basic_decomposition: minimum value of jpi must be >= ', 2*i2hls |
|---|
| 759 | WRITE(ctmp2,*) ' We have ', MINVAL(klci) |
|---|
| 760 | CALL ctl_stop( 'STOP', ctmp1, ctmp2 ) |
|---|
| 761 | ENDIF |
|---|
| 762 | IF( jperio == 3 .OR. jperio == 4 .OR. jperio == 5 .OR. jperio == 6 ) THEN |
|---|
| 763 | ! minimize the size of the last row to compensate for the north pole folding coast |
|---|
| 764 | IF( jperio == 3 .OR. jperio == 4 ) ijpjmin = 2+3*khls ! V and F folding must be outside of southern halos |
|---|
| 765 | IF( jperio == 5 .OR. jperio == 6 ) ijpjmin = 1+3*khls ! V and F folding must be outside of southern halos |
|---|
| 766 | irm = knbj - irestj ! total number of lines to be removed |
|---|
| 767 | klcj(:,knbj) = MAX( ijpjmin, kjmax-irm ) ! we must have jpj >= ijpjmin in the last row |
|---|
| 768 | irm = irm - ( kjmax - klcj(1,knbj) ) ! remaining number of lines to remove |
|---|
| 769 | irestj = knbj - 1 - irm |
|---|
| 770 | klcj(:, irestj+1:knbj-1) = kjmax-1 |
|---|
| 771 | ELSE |
|---|
| 772 | klcj(:, irestj+1:knbj ) = kjmax-1 |
|---|
| 773 | ENDIF |
|---|
| 774 | klcj(:,1:irestj) = kjmax |
|---|
| 775 | IF( MINVAL(klcj) < 2*i2hls ) THEN |
|---|
| 776 | WRITE(ctmp1,*) ' basic_decomposition: minimum value of jpj must be >= ', 2*i2hls |
|---|
| 777 | WRITE(ctmp2,*) ' We have ', MINVAL(klcj) |
|---|
| 778 | CALL ctl_stop( 'STOP', ctmp1, ctmp2 ) |
|---|
| 779 | ENDIF |
|---|
| 780 | #endif |
|---|
| 781 | |
|---|
| 782 | ! 2. Index arrays for subdomains |
|---|
| 783 | ! ------------------------------- |
|---|
| 784 | kimppt(:,:) = 1 |
|---|
| 785 | kjmppt(:,:) = 1 |
|---|
| 786 | ! |
|---|
| 787 | IF( knbi > 1 ) THEN |
|---|
| 788 | DO jj = 1, knbj |
|---|
| 789 | DO ji = 2, knbi |
|---|
| 790 | kimppt(ji,jj) = kimppt(ji-1,jj) + klci(ji-1,jj) - 2 * nn_hls |
|---|
| 791 | END DO |
|---|
| 792 | END DO |
|---|
| 793 | ENDIF |
|---|
| 794 | ! |
|---|
| 795 | IF( knbj > 1 )THEN |
|---|
| 796 | DO jj = 2, knbj |
|---|
| 797 | DO ji = 1, knbi |
|---|
| 798 | kjmppt(ji,jj) = kjmppt(ji,jj-1) + klcj(ji,jj-1) - 2 * nn_hls |
|---|
| 799 | END DO |
|---|
| 800 | END DO |
|---|
| 801 | ENDIF |
|---|
| 802 | |
|---|
| 803 | END SUBROUTINE basic_decomposition |
|---|
| 804 | |
|---|
| 805 | |
|---|
| 806 | SUBROUTINE bestpartition( knbij, knbi, knbj, knbcnt, ldlist ) |
|---|
| 807 | !!---------------------------------------------------------------------- |
|---|
| 808 | !! *** ROUTINE bestpartition *** |
|---|
| 809 | !! |
|---|
| 810 | !! ** Purpose : |
|---|
| 811 | !! |
|---|
| 812 | !! ** Method : |
|---|
| 813 | !!---------------------------------------------------------------------- |
|---|
| 814 | INTEGER, INTENT(in ) :: knbij ! total number if subdomains (knbi*knbj) |
|---|
| 815 | INTEGER, OPTIONAL, INTENT( out) :: knbi, knbj ! number if subdomains along i and j (knbi and knbj) |
|---|
| 816 | INTEGER, OPTIONAL, INTENT( out) :: knbcnt ! number of land subdomains |
|---|
| 817 | LOGICAL, OPTIONAL, INTENT(in ) :: ldlist ! .true.: print the list the best domain decompositions (with land) |
|---|
| 818 | ! |
|---|
| 819 | INTEGER :: ji, jj, ii, iitarget |
|---|
| 820 | INTEGER :: iszitst, iszjtst |
|---|
| 821 | INTEGER :: isziref, iszjref |
|---|
| 822 | INTEGER :: inbij, iszij |
|---|
| 823 | INTEGER :: inbimax, inbjmax, inbijmax, inbijold |
|---|
| 824 | INTEGER :: isz0, isz1 |
|---|
| 825 | INTEGER, DIMENSION( :), ALLOCATABLE :: indexok |
|---|
| 826 | INTEGER, DIMENSION( :), ALLOCATABLE :: inbi0, inbj0, inbij0 ! number of subdomains along i,j |
|---|
| 827 | INTEGER, DIMENSION( :), ALLOCATABLE :: iszi0, iszj0, iszij0 ! max size of the subdomains along i,j |
|---|
| 828 | INTEGER, DIMENSION( :), ALLOCATABLE :: inbi1, inbj1, inbij1 ! number of subdomains along i,j |
|---|
| 829 | INTEGER, DIMENSION( :), ALLOCATABLE :: iszi1, iszj1, iszij1 ! max size of the subdomains along i,j |
|---|
| 830 | LOGICAL :: llist |
|---|
| 831 | LOGICAL, DIMENSION(:,:), ALLOCATABLE :: llmsk2d ! max size of the subdomains along i,j |
|---|
| 832 | LOGICAL, DIMENSION(:,:), ALLOCATABLE :: llisoce ! - - |
|---|
| 833 | REAL(wp):: zpropland |
|---|
| 834 | !!---------------------------------------------------------------------- |
|---|
| 835 | ! |
|---|
| 836 | llist = .FALSE. |
|---|
| 837 | IF( PRESENT(ldlist) ) llist = ldlist |
|---|
| 838 | |
|---|
| 839 | CALL mpp_init_landprop( zpropland ) ! get the proportion of land point over the gloal domain |
|---|
| 840 | inbij = NINT( REAL(knbij, wp) / ( 1.0 - zpropland ) ) ! define the largest possible value for jpni*jpnj |
|---|
| 841 | ! |
|---|
| 842 | IF( llist ) THEN ; inbijmax = inbij*2 |
|---|
| 843 | ELSE ; inbijmax = inbij |
|---|
| 844 | ENDIF |
|---|
| 845 | ! |
|---|
| 846 | ALLOCATE(inbi0(inbijmax),inbj0(inbijmax),iszi0(inbijmax),iszj0(inbijmax)) |
|---|
| 847 | ! |
|---|
| 848 | inbimax = 0 |
|---|
| 849 | inbjmax = 0 |
|---|
| 850 | isziref = jpiglo*jpjglo+1 |
|---|
| 851 | iszjref = jpiglo*jpjglo+1 |
|---|
| 852 | ! |
|---|
| 853 | ! get the list of knbi that gives a smaller jpimax than knbi-1 |
|---|
| 854 | ! get the list of knbj that gives a smaller jpjmax than knbj-1 |
|---|
| 855 | DO ji = 1, inbijmax |
|---|
| 856 | #if defined key_nemocice_decomp |
|---|
| 857 | iszitst = ( nx_global+2-2*nn_hls + (ji-1) ) / ji + 2*nn_hls ! first dim. |
|---|
| 858 | #else |
|---|
| 859 | iszitst = ( Ni0glo + (ji-1) ) / ji + 2*nn_hls |
|---|
| 860 | #endif |
|---|
| 861 | IF( iszitst < isziref ) THEN |
|---|
| 862 | isziref = iszitst |
|---|
| 863 | inbimax = inbimax + 1 |
|---|
| 864 | inbi0(inbimax) = ji |
|---|
| 865 | iszi0(inbimax) = isziref |
|---|
| 866 | ENDIF |
|---|
| 867 | #if defined key_nemocice_decomp |
|---|
| 868 | iszjtst = ( ny_global+2-2*nn_hls + (ji-1) ) / ji + 2*nn_hls ! first dim. |
|---|
| 869 | #else |
|---|
| 870 | iszjtst = ( Nj0glo + (ji-1) ) / ji + 2*nn_hls |
|---|
| 871 | #endif |
|---|
| 872 | IF( iszjtst < iszjref ) THEN |
|---|
| 873 | iszjref = iszjtst |
|---|
| 874 | inbjmax = inbjmax + 1 |
|---|
| 875 | inbj0(inbjmax) = ji |
|---|
| 876 | iszj0(inbjmax) = iszjref |
|---|
| 877 | ENDIF |
|---|
| 878 | END DO |
|---|
| 879 | |
|---|
| 880 | ! combine these 2 lists to get all possible knbi*knbj < inbijmax |
|---|
| 881 | ALLOCATE( llmsk2d(inbimax,inbjmax) ) |
|---|
| 882 | DO jj = 1, inbjmax |
|---|
| 883 | DO ji = 1, inbimax |
|---|
| 884 | IF ( inbi0(ji) * inbj0(jj) <= inbijmax ) THEN ; llmsk2d(ji,jj) = .TRUE. |
|---|
| 885 | ELSE ; llmsk2d(ji,jj) = .FALSE. |
|---|
| 886 | ENDIF |
|---|
| 887 | END DO |
|---|
| 888 | END DO |
|---|
| 889 | isz1 = COUNT(llmsk2d) |
|---|
| 890 | ALLOCATE( inbi1(isz1), inbj1(isz1), iszi1(isz1), iszj1(isz1) ) |
|---|
| 891 | ii = 0 |
|---|
| 892 | DO jj = 1, inbjmax |
|---|
| 893 | DO ji = 1, inbimax |
|---|
| 894 | IF( llmsk2d(ji,jj) .EQV. .TRUE. ) THEN |
|---|
| 895 | ii = ii + 1 |
|---|
| 896 | inbi1(ii) = inbi0(ji) |
|---|
| 897 | inbj1(ii) = inbj0(jj) |
|---|
| 898 | iszi1(ii) = iszi0(ji) |
|---|
| 899 | iszj1(ii) = iszj0(jj) |
|---|
| 900 | END IF |
|---|
| 901 | END DO |
|---|
| 902 | END DO |
|---|
| 903 | DEALLOCATE( inbi0, inbj0, iszi0, iszj0 ) |
|---|
| 904 | DEALLOCATE( llmsk2d ) |
|---|
| 905 | |
|---|
| 906 | ALLOCATE( inbij1(isz1), iszij1(isz1) ) |
|---|
| 907 | inbij1(:) = inbi1(:) * inbj1(:) |
|---|
| 908 | iszij1(:) = iszi1(:) * iszj1(:) |
|---|
| 909 | |
|---|
| 910 | ! if therr is no land and no print |
|---|
| 911 | IF( .NOT. llist .AND. numbot == -1 .AND. numbdy == -1 ) THEN |
|---|
| 912 | ! get the smaller partition which gives the smallest subdomain size |
|---|
| 913 | ii = MINLOC(inbij1, mask = iszij1 == MINVAL(iszij1), dim = 1) |
|---|
| 914 | knbi = inbi1(ii) |
|---|
| 915 | knbj = inbj1(ii) |
|---|
| 916 | IF(PRESENT(knbcnt)) knbcnt = 0 |
|---|
| 917 | DEALLOCATE( inbi1, inbj1, inbij1, iszi1, iszj1, iszij1 ) |
|---|
| 918 | RETURN |
|---|
| 919 | ENDIF |
|---|
| 920 | |
|---|
| 921 | ! extract only the partitions which reduce the subdomain size in comparison with smaller partitions |
|---|
| 922 | ALLOCATE( indexok(isz1) ) ! to store indices of the best partitions |
|---|
| 923 | isz0 = 0 ! number of best partitions |
|---|
| 924 | inbij = 1 ! start with the min value of inbij1 => 1 |
|---|
| 925 | iszij = jpiglo*jpjglo+1 ! default: larger than global domain |
|---|
| 926 | DO WHILE( inbij <= inbijmax ) ! if we did not reach the max of inbij1 |
|---|
| 927 | ii = MINLOC(iszij1, mask = inbij1 == inbij, dim = 1) ! warning: send back the first occurence if multiple results |
|---|
| 928 | IF ( iszij1(ii) < iszij ) THEN |
|---|
| 929 | isz0 = isz0 + 1 |
|---|
| 930 | indexok(isz0) = ii |
|---|
| 931 | iszij = iszij1(ii) |
|---|
| 932 | ENDIF |
|---|
| 933 | inbij = MINVAL(inbij1, mask = inbij1 > inbij) ! warning: return largest integer value if mask = .false. everywhere |
|---|
| 934 | END DO |
|---|
| 935 | DEALLOCATE( inbij1, iszij1 ) |
|---|
| 936 | |
|---|
| 937 | ! keep only the best partitions (sorted by increasing order of subdomains number and decreassing subdomain size) |
|---|
| 938 | ALLOCATE( inbi0(isz0), inbj0(isz0), iszi0(isz0), iszj0(isz0) ) |
|---|
| 939 | DO ji = 1, isz0 |
|---|
| 940 | ii = indexok(ji) |
|---|
| 941 | inbi0(ji) = inbi1(ii) |
|---|
| 942 | inbj0(ji) = inbj1(ii) |
|---|
| 943 | iszi0(ji) = iszi1(ii) |
|---|
| 944 | iszj0(ji) = iszj1(ii) |
|---|
| 945 | END DO |
|---|
| 946 | DEALLOCATE( indexok, inbi1, inbj1, iszi1, iszj1 ) |
|---|
| 947 | |
|---|
| 948 | IF( llist ) THEN |
|---|
| 949 | IF(lwp) THEN |
|---|
| 950 | WRITE(numout,*) |
|---|
| 951 | WRITE(numout,*) ' For your information:' |
|---|
| 952 | WRITE(numout,*) ' list of the best partitions including land supression' |
|---|
| 953 | WRITE(numout,*) ' -----------------------------------------------------' |
|---|
| 954 | WRITE(numout,*) |
|---|
| 955 | END IF |
|---|
| 956 | ji = isz0 ! initialization with the largest value |
|---|
| 957 | ALLOCATE( llisoce(inbi0(ji), inbj0(ji)) ) |
|---|
| 958 | CALL is_ocean( inbi0(ji), inbj0(ji), llisoce ) ! Warning: must be call by all cores (call mpp_sum) |
|---|
| 959 | inbijold = COUNT(llisoce) |
|---|
| 960 | DEALLOCATE( llisoce ) |
|---|
| 961 | DO ji =isz0-1,1,-1 |
|---|
| 962 | ALLOCATE( llisoce(inbi0(ji), inbj0(ji)) ) |
|---|
| 963 | CALL is_ocean( inbi0(ji), inbj0(ji), llisoce ) ! Warning: must be call by all cores (call mpp_sum) |
|---|
| 964 | inbij = COUNT(llisoce) |
|---|
| 965 | DEALLOCATE( llisoce ) |
|---|
| 966 | IF(lwp .AND. inbij < inbijold) THEN |
|---|
| 967 | WRITE(numout,'(a, i6, a, i6, a, f4.1, a, i9, a, i6, a, i6, a)') & |
|---|
| 968 | & 'nb_cores oce: ', inbij, ', land domains excluded: ', inbi0(ji)*inbj0(ji) - inbij, & |
|---|
| 969 | & ' (', REAL(inbi0(ji)*inbj0(ji) - inbij,wp) / REAL(inbi0(ji)*inbj0(ji),wp) *100., & |
|---|
| 970 | & '%), largest oce domain: ', iszi0(ji)*iszj0(ji), ' ( ', iszi0(ji),' x ', iszj0(ji), ' )' |
|---|
| 971 | inbijold = inbij |
|---|
| 972 | END IF |
|---|
| 973 | END DO |
|---|
| 974 | DEALLOCATE( inbi0, inbj0, iszi0, iszj0 ) |
|---|
| 975 | IF(lwp) THEN |
|---|
| 976 | WRITE(numout,*) |
|---|
| 977 | WRITE(numout,*) ' -----------------------------------------------------------' |
|---|
| 978 | ENDIF |
|---|
| 979 | CALL mppsync |
|---|
| 980 | CALL mppstop( ld_abort = .TRUE. ) |
|---|
| 981 | ENDIF |
|---|
| 982 | |
|---|
| 983 | DEALLOCATE( iszi0, iszj0 ) |
|---|
| 984 | inbij = inbijmax + 1 ! default: larger than possible |
|---|
| 985 | ii = isz0+1 ! start from the end of the list (smaller subdomains) |
|---|
| 986 | DO WHILE( inbij > knbij ) ! while the number of ocean subdomains exceed the number of procs |
|---|
| 987 | ii = ii -1 |
|---|
| 988 | ALLOCATE( llisoce(inbi0(ii), inbj0(ii)) ) |
|---|
| 989 | CALL is_ocean( inbi0(ii), inbj0(ii), llisoce ) ! must be done by all core |
|---|
| 990 | inbij = COUNT(llisoce) |
|---|
| 991 | DEALLOCATE( llisoce ) |
|---|
| 992 | END DO |
|---|
| 993 | knbi = inbi0(ii) |
|---|
| 994 | knbj = inbj0(ii) |
|---|
| 995 | IF(PRESENT(knbcnt)) knbcnt = knbi * knbj - inbij |
|---|
| 996 | DEALLOCATE( inbi0, inbj0 ) |
|---|
| 997 | ! |
|---|
| 998 | END SUBROUTINE bestpartition |
|---|
| 999 | |
|---|
| 1000 | |
|---|
| 1001 | SUBROUTINE mpp_init_landprop( propland ) |
|---|
| 1002 | !!---------------------------------------------------------------------- |
|---|
| 1003 | !! *** ROUTINE mpp_init_landprop *** |
|---|
| 1004 | !! |
|---|
| 1005 | !! ** Purpose : the the proportion of land points in the surface land-sea mask |
|---|
| 1006 | !! |
|---|
| 1007 | !! ** Method : read iproc strips (of length Ni0glo) of the land-sea mask |
|---|
| 1008 | !!---------------------------------------------------------------------- |
|---|
| 1009 | REAL(wp), INTENT( out) :: propland ! proportion of land points in the global domain (between 0 and 1) |
|---|
| 1010 | ! |
|---|
| 1011 | INTEGER, DIMENSION(jpni*jpnj) :: kusedom_1d |
|---|
| 1012 | INTEGER :: inboce, iarea |
|---|
| 1013 | INTEGER :: iproc, idiv, ijsz |
|---|
| 1014 | INTEGER :: ijstr |
|---|
| 1015 | LOGICAL, ALLOCATABLE, DIMENSION(:,:) :: lloce |
|---|
| 1016 | !!---------------------------------------------------------------------- |
|---|
| 1017 | ! do nothing if there is no land-sea mask |
|---|
| 1018 | IF( numbot == -1 .and. numbdy == -1 ) THEN |
|---|
| 1019 | propland = 0. |
|---|
| 1020 | RETURN |
|---|
| 1021 | ENDIF |
|---|
| 1022 | |
|---|
| 1023 | ! number of processes reading the bathymetry file |
|---|
| 1024 | iproc = MINVAL( (/mppsize, Nj0glo/2, 100/) ) ! read a least 2 lines, no more that 100 processes reading at the same time |
|---|
| 1025 | |
|---|
| 1026 | ! we want to read iproc strips of the land-sea mask. -> pick up iproc processes every idiv processes starting at 1 |
|---|
| 1027 | IF( iproc == 1 ) THEN ; idiv = mppsize |
|---|
| 1028 | ELSE ; idiv = ( mppsize - 1 ) / ( iproc - 1 ) |
|---|
| 1029 | ENDIF |
|---|
| 1030 | |
|---|
| 1031 | iarea = (narea-1)/idiv ! involed process number (starting counting at 0) |
|---|
| 1032 | IF( MOD( narea-1, idiv ) == 0 .AND. iarea < iproc ) THEN ! beware idiv can be = to 1 |
|---|
| 1033 | ! |
|---|
| 1034 | ijsz = Nj0glo / iproc ! width of the stripe to read |
|---|
| 1035 | IF( iarea < MOD(Nj0glo,iproc) ) ijsz = ijsz + 1 |
|---|
| 1036 | ijstr = iarea*(Nj0glo/iproc) + MIN(iarea, MOD(Nj0glo,iproc)) + 1 ! starting j position of the reading |
|---|
| 1037 | ! |
|---|
| 1038 | ALLOCATE( lloce(Ni0glo, ijsz) ) ! allocate the strip |
|---|
| 1039 | CALL readbot_strip( ijstr, ijsz, lloce ) |
|---|
| 1040 | inboce = COUNT(lloce) ! number of ocean point in the stripe |
|---|
| 1041 | DEALLOCATE(lloce) |
|---|
| 1042 | ! |
|---|
| 1043 | ELSE |
|---|
| 1044 | inboce = 0 |
|---|
| 1045 | ENDIF |
|---|
| 1046 | CALL mpp_sum( 'mppini', inboce ) ! total number of ocean points over the global domain |
|---|
| 1047 | ! |
|---|
| 1048 | propland = REAL( Ni0glo*Nj0glo - inboce, wp ) / REAL( Ni0glo*Nj0glo, wp ) |
|---|
| 1049 | ! |
|---|
| 1050 | END SUBROUTINE mpp_init_landprop |
|---|
| 1051 | |
|---|
| 1052 | |
|---|
| 1053 | SUBROUTINE is_ocean( knbi, knbj, ldisoce ) |
|---|
| 1054 | !!---------------------------------------------------------------------- |
|---|
| 1055 | !! *** ROUTINE mpp_init_nboce *** |
|---|
| 1056 | !! |
|---|
| 1057 | !! ** Purpose : check for a mpi domain decomposition knbi x knbj which |
|---|
| 1058 | !! subdomains contain at least 1 ocean point |
|---|
| 1059 | !! |
|---|
| 1060 | !! ** Method : read knbj strips (of length Ni0glo) of the land-sea mask |
|---|
| 1061 | !!---------------------------------------------------------------------- |
|---|
| 1062 | INTEGER, INTENT(in ) :: knbi, knbj ! domain decomposition |
|---|
| 1063 | LOGICAL, DIMENSION(knbi,knbj), INTENT( out) :: ldisoce ! .true. if a sub domain constains 1 ocean point |
|---|
| 1064 | ! |
|---|
| 1065 | INTEGER, DIMENSION(knbi,knbj) :: inboce ! number oce oce pint in each mpi subdomain |
|---|
| 1066 | INTEGER, DIMENSION(knbi*knbj) :: inboce_1d |
|---|
| 1067 | INTEGER :: idiv, iimax, ijmax, iarea |
|---|
| 1068 | INTEGER :: ji, jn |
|---|
| 1069 | LOGICAL, ALLOCATABLE, DIMENSION(:,:) :: lloce ! lloce(i,j) = .true. if the point (i,j) is ocean |
|---|
| 1070 | INTEGER, ALLOCATABLE, DIMENSION(:,:) :: iimppt, ijpi |
|---|
| 1071 | INTEGER, ALLOCATABLE, DIMENSION(:,:) :: ijmppt, ijpj |
|---|
| 1072 | !!---------------------------------------------------------------------- |
|---|
| 1073 | ! do nothing if there is no land-sea mask |
|---|
| 1074 | IF( numbot == -1 .AND. numbdy == -1 ) THEN |
|---|
| 1075 | ldisoce(:,:) = .TRUE. |
|---|
| 1076 | RETURN |
|---|
| 1077 | ENDIF |
|---|
| 1078 | |
|---|
| 1079 | ! we want to read knbj strips of the land-sea mask. -> pick up knbj processes every idiv processes starting at 1 |
|---|
| 1080 | IF ( knbj == 1 ) THEN ; idiv = mppsize |
|---|
| 1081 | ELSE IF ( mppsize < knbj ) THEN ; idiv = 1 |
|---|
| 1082 | ELSE ; idiv = ( mppsize - 1 ) / ( knbj - 1 ) |
|---|
| 1083 | ENDIF |
|---|
| 1084 | inboce(:,:) = 0 ! default no ocean point found |
|---|
| 1085 | |
|---|
| 1086 | DO jn = 0, (knbj-1)/mppsize ! if mppsize < knbj : more strips than mpi processes (because of potential land domains) |
|---|
| 1087 | ! |
|---|
| 1088 | iarea = (narea-1)/idiv + jn * mppsize ! involed process number (starting counting at 0) |
|---|
| 1089 | IF( MOD( narea-1, idiv ) == 0 .AND. iarea < knbj ) THEN ! beware idiv can be = to 1 |
|---|
| 1090 | ! |
|---|
| 1091 | ALLOCATE( iimppt(knbi,knbj), ijmppt(knbi,knbj), ijpi(knbi,knbj), ijpj(knbi,knbj) ) |
|---|
| 1092 | CALL basic_decomposition( Ni0glo, Nj0glo, 0, knbi, knbj, iimax, ijmax, iimppt, ijmppt, ijpi, ijpj ) |
|---|
| 1093 | ! |
|---|
| 1094 | ALLOCATE( lloce(Ni0glo, ijpj(1,iarea+1)) ) ! allocate the strip |
|---|
| 1095 | CALL readbot_strip( ijmppt(1,iarea+1), ijpj(1,iarea+1), lloce ) ! read the strip |
|---|
| 1096 | DO ji = 1, knbi |
|---|
| 1097 | inboce(ji,iarea+1) = COUNT( lloce(iimppt(ji,1):iimppt(ji,1)+ijpi(ji,1)-1,:) ) ! number of ocean point in subdomain |
|---|
| 1098 | END DO |
|---|
| 1099 | ! |
|---|
| 1100 | DEALLOCATE(lloce) |
|---|
| 1101 | DEALLOCATE(iimppt, ijmppt, ijpi, ijpj) |
|---|
| 1102 | ! |
|---|
| 1103 | ENDIF |
|---|
| 1104 | END DO |
|---|
| 1105 | |
|---|
| 1106 | inboce_1d = RESHAPE(inboce, (/ knbi*knbj /)) |
|---|
| 1107 | CALL mpp_sum( 'mppini', inboce_1d ) |
|---|
| 1108 | inboce = RESHAPE(inboce_1d, (/knbi, knbj/)) |
|---|
| 1109 | ldisoce(:,:) = inboce(:,:) /= 0 |
|---|
| 1110 | ! |
|---|
| 1111 | END SUBROUTINE is_ocean |
|---|
| 1112 | |
|---|
| 1113 | |
|---|
| 1114 | SUBROUTINE readbot_strip( kjstr, kjcnt, ldoce ) |
|---|
| 1115 | !!---------------------------------------------------------------------- |
|---|
| 1116 | !! *** ROUTINE readbot_strip *** |
|---|
| 1117 | !! |
|---|
| 1118 | !! ** Purpose : Read relevant bathymetric information in order to |
|---|
| 1119 | !! provide a land/sea mask used for the elimination |
|---|
| 1120 | !! of land domains, in an mpp computation. |
|---|
| 1121 | !! |
|---|
| 1122 | !! ** Method : read stipe of size (Ni0glo,...) |
|---|
| 1123 | !!---------------------------------------------------------------------- |
|---|
| 1124 | INTEGER , INTENT(in ) :: kjstr ! starting j position of the reading |
|---|
| 1125 | INTEGER , INTENT(in ) :: kjcnt ! number of lines to read |
|---|
| 1126 | LOGICAL, DIMENSION(Ni0glo,kjcnt), INTENT( out) :: ldoce ! ldoce(i,j) = .true. if the point (i,j) is ocean |
|---|
| 1127 | ! |
|---|
| 1128 | INTEGER :: inumsave ! local logical unit |
|---|
| 1129 | REAL(wp), DIMENSION(Ni0glo,kjcnt) :: zbot, zbdy |
|---|
| 1130 | !!---------------------------------------------------------------------- |
|---|
| 1131 | ! |
|---|
| 1132 | inumsave = numout ; numout = numnul ! redirect all print to /dev/null |
|---|
| 1133 | ! |
|---|
| 1134 | IF( numbot /= -1 ) THEN |
|---|
| 1135 | CALL iom_get( numbot, jpdom_unknown, 'bottom_level', zbot, kstart = (/1,kjstr/), kcount = (/Ni0glo, kjcnt/) ) |
|---|
| 1136 | ELSE |
|---|
| 1137 | zbot(:,:) = 1._wp ! put a non-null value |
|---|
| 1138 | ENDIF |
|---|
| 1139 | ! |
|---|
| 1140 | IF( numbdy /= -1 ) THEN ! Adjust with bdy_msk if it exists |
|---|
| 1141 | CALL iom_get ( numbdy, jpdom_unknown, 'bdy_msk', zbdy, kstart = (/1,kjstr/), kcount = (/Ni0glo, kjcnt/) ) |
|---|
| 1142 | zbot(:,:) = zbot(:,:) * zbdy(:,:) |
|---|
| 1143 | ENDIF |
|---|
| 1144 | ! |
|---|
| 1145 | ldoce(:,:) = zbot(:,:) > 0._wp |
|---|
| 1146 | numout = inumsave |
|---|
| 1147 | ! |
|---|
| 1148 | END SUBROUTINE readbot_strip |
|---|
| 1149 | |
|---|
| 1150 | |
|---|
| 1151 | SUBROUTINE init_ioipsl |
|---|
| 1152 | !!---------------------------------------------------------------------- |
|---|
| 1153 | !! *** ROUTINE init_ioipsl *** |
|---|
| 1154 | !! |
|---|
| 1155 | !! ** Purpose : |
|---|
| 1156 | !! |
|---|
| 1157 | !! ** Method : |
|---|
| 1158 | !! |
|---|
| 1159 | !! History : |
|---|
| 1160 | !! 9.0 ! 04-03 (G. Madec ) MPP-IOIPSL |
|---|
| 1161 | !! " " ! 08-12 (A. Coward) addition in case of jpni*jpnj < jpnij |
|---|
| 1162 | !!---------------------------------------------------------------------- |
|---|
| 1163 | INTEGER, DIMENSION(2) :: iglo, iloc, iabsf, iabsl, ihals, ihale, idid |
|---|
| 1164 | !!---------------------------------------------------------------------- |
|---|
| 1165 | |
|---|
| 1166 | ! The domain is split only horizontally along i- or/and j- direction |
|---|
| 1167 | ! So we need at the most only 1D arrays with 2 elements. |
|---|
| 1168 | ! Set idompar values equivalent to the jpdom_local_noextra definition |
|---|
| 1169 | ! used in IOM. This works even if jpnij .ne. jpni*jpnj. |
|---|
| 1170 | iglo(1) = jpiglo |
|---|
| 1171 | iglo(2) = jpjglo |
|---|
| 1172 | iloc(1) = jpi |
|---|
| 1173 | iloc(2) = jpj |
|---|
| 1174 | iabsf(1) = nimppt(narea) |
|---|
| 1175 | iabsf(2) = njmppt(narea) |
|---|
| 1176 | iabsl(:) = iabsf(:) + iloc(:) - 1 |
|---|
| 1177 | ihals(1) = Nis0 - 1 |
|---|
| 1178 | ihals(2) = Njs0 - 1 |
|---|
| 1179 | ihale(1) = jpi - Nie0 |
|---|
| 1180 | ihale(2) = jpj - Nje0 |
|---|
| 1181 | idid(1) = 1 |
|---|
| 1182 | idid(2) = 2 |
|---|
| 1183 | |
|---|
| 1184 | IF(lwp) THEN |
|---|
| 1185 | WRITE(numout,*) |
|---|
| 1186 | WRITE(numout,*) 'mpp init_ioipsl : iloc = ', iloc (1), iloc (2) |
|---|
| 1187 | WRITE(numout,*) '~~~~~~~~~~~~~~~ iabsf = ', iabsf(1), iabsf(2) |
|---|
| 1188 | WRITE(numout,*) ' ihals = ', ihals(1), ihals(2) |
|---|
| 1189 | WRITE(numout,*) ' ihale = ', ihale(1), ihale(2) |
|---|
| 1190 | ENDIF |
|---|
| 1191 | ! |
|---|
| 1192 | CALL flio_dom_set ( jpnij, nproc, idid, iglo, iloc, iabsf, iabsl, ihals, ihale, 'BOX', nidom) |
|---|
| 1193 | ! |
|---|
| 1194 | END SUBROUTINE init_ioipsl |
|---|
| 1195 | |
|---|
| 1196 | |
|---|
| 1197 | SUBROUTINE init_nfdcom |
|---|
| 1198 | !!---------------------------------------------------------------------- |
|---|
| 1199 | !! *** ROUTINE init_nfdcom *** |
|---|
| 1200 | !! ** Purpose : Setup for north fold exchanges with explicit |
|---|
| 1201 | !! point-to-point messaging |
|---|
| 1202 | !! |
|---|
| 1203 | !! ** Method : Initialization of the northern neighbours lists. |
|---|
| 1204 | !!---------------------------------------------------------------------- |
|---|
| 1205 | !! 1.0 ! 2011-10 (A. C. Coward, NOCS & J. Donners, PRACE) |
|---|
| 1206 | !! 2.0 ! 2013-06 Setup avoiding MPI communication (I. Epicoco, S. Mocavero, CMCC) |
|---|
| 1207 | !!---------------------------------------------------------------------- |
|---|
| 1208 | INTEGER :: sxM, dxM, sxT, dxT, jn |
|---|
| 1209 | INTEGER :: njmppmax |
|---|
| 1210 | !!---------------------------------------------------------------------- |
|---|
| 1211 | ! |
|---|
| 1212 | njmppmax = MAXVAL( njmppt ) |
|---|
| 1213 | ! |
|---|
| 1214 | !initializes the north-fold communication variables |
|---|
| 1215 | isendto(:) = 0 |
|---|
| 1216 | nsndto = 0 |
|---|
| 1217 | ! |
|---|
| 1218 | IF ( njmpp == njmppmax ) THEN ! if I am a process in the north |
|---|
| 1219 | ! |
|---|
| 1220 | !sxM is the first point (in the global domain) needed to compute the north-fold for the current process |
|---|
| 1221 | sxM = jpiglo - nimppt(narea) - jpiall(narea) + 1 |
|---|
| 1222 | !dxM is the last point (in the global domain) needed to compute the north-fold for the current process |
|---|
| 1223 | dxM = jpiglo - nimppt(narea) + 2 |
|---|
| 1224 | ! |
|---|
| 1225 | ! loop over the other north-fold processes to find the processes |
|---|
| 1226 | ! managing the points belonging to the sxT-dxT range |
|---|
| 1227 | ! |
|---|
| 1228 | DO jn = 1, jpni |
|---|
| 1229 | ! |
|---|
| 1230 | sxT = nfiimpp(jn, jpnj) ! sxT = 1st point (in the global domain) of the jn process |
|---|
| 1231 | dxT = nfiimpp(jn, jpnj) + nfijpit(jn, jpnj) - 1 ! dxT = last point (in the global domain) of the jn process |
|---|
| 1232 | ! |
|---|
| 1233 | IF ( sxT < sxM .AND. sxM < dxT ) THEN |
|---|
| 1234 | nsndto = nsndto + 1 |
|---|
| 1235 | isendto(nsndto) = jn |
|---|
| 1236 | ELSEIF( sxM <= sxT .AND. dxM >= dxT ) THEN |
|---|
| 1237 | nsndto = nsndto + 1 |
|---|
| 1238 | isendto(nsndto) = jn |
|---|
| 1239 | ELSEIF( dxM < dxT .AND. sxT < dxM ) THEN |
|---|
| 1240 | nsndto = nsndto + 1 |
|---|
| 1241 | isendto(nsndto) = jn |
|---|
| 1242 | ENDIF |
|---|
| 1243 | ! |
|---|
| 1244 | END DO |
|---|
| 1245 | ! |
|---|
| 1246 | ENDIF |
|---|
| 1247 | l_north_nogather = .TRUE. |
|---|
| 1248 | ! |
|---|
| 1249 | END SUBROUTINE init_nfdcom |
|---|
| 1250 | |
|---|
| 1251 | #endif |
|---|
| 1252 | |
|---|
| 1253 | SUBROUTINE init_doloop |
|---|
| 1254 | !!---------------------------------------------------------------------- |
|---|
| 1255 | !! *** ROUTINE init_doloop *** |
|---|
| 1256 | !! |
|---|
| 1257 | !! ** Purpose : set the starting/ending indices of DO-loop |
|---|
| 1258 | !! These indices are used in do_loop_substitute.h90 |
|---|
| 1259 | !!---------------------------------------------------------------------- |
|---|
| 1260 | ! |
|---|
| 1261 | IF( nn_hls == 1 ) THEN !* halo size of 1 |
|---|
| 1262 | ! |
|---|
| 1263 | Nis0 = 2 ; Nis1 = 1 ; Nis1nxt2 = Nis0 ; Nis2 = Nis1 |
|---|
| 1264 | Njs0 = 2 ; Njs1 = 1 ; Njs1nxt2 = Njs0 ; Njs2 = Njs1 |
|---|
| 1265 | ! |
|---|
| 1266 | Nie0 = jpi-1 ; Nie1 = jpi ; Nie1nxt2 = Nie0 ; Nie2 = Nie1 |
|---|
| 1267 | Nje0 = jpj-1 ; Nje1 = jpj ; Nje1nxt2 = Nje0 ; Nje2 = Nje1 |
|---|
| 1268 | ! |
|---|
| 1269 | ELSEIF( nn_hls == 2 ) THEN !* halo size of 2 |
|---|
| 1270 | ! |
|---|
| 1271 | Nis0 = 3 ; Nis1 = 2 ; Nis1nxt2 = Nis1 ; Nis2 = 1 |
|---|
| 1272 | Njs0 = 3 ; Njs1 = 2 ; Njs1nxt2 = Njs1 ; Njs2 = 1 |
|---|
| 1273 | ! |
|---|
| 1274 | Nie0 = jpi-2 ; Nie1 = jpi-1 ; Nie1nxt2 = Nie1 ; Nie2 = jpi |
|---|
| 1275 | Nje0 = jpj-2 ; Nje1 = jpj-1 ; Nje1nxt2 = Nje1 ; Nje2 = jpj |
|---|
| 1276 | ! |
|---|
| 1277 | ELSE !* unexpected halo size |
|---|
| 1278 | CALL ctl_stop( 'STOP', 'ini_mpp: wrong value of halo size : nn_hls= 1 or 2 only !') |
|---|
| 1279 | ENDIF |
|---|
| 1280 | ! |
|---|
| 1281 | Ni_0 = Nie0 - Nis0 + 1 |
|---|
| 1282 | Nj_0 = Nje0 - Njs0 + 1 |
|---|
| 1283 | Ni_1 = Nie1 - Nis1 + 1 |
|---|
| 1284 | Nj_1 = Nje1 - Njs1 + 1 |
|---|
| 1285 | Ni_2 = Nie2 - Nis2 + 1 |
|---|
| 1286 | Nj_2 = Nje2 - Njs2 + 1 |
|---|
| 1287 | ! |
|---|
| 1288 | END SUBROUTINE init_doloop |
|---|
| 1289 | |
|---|
| 1290 | !!====================================================================== |
|---|
| 1291 | END MODULE mppini |
|---|
