1 | MODULE lib_mpp |
---|
2 | !!====================================================================== |
---|
3 | !! *** MODULE lib_mpp *** |
---|
4 | !! Ocean numerics: massively parallel processing library |
---|
5 | !!===================================================================== |
---|
6 | !! History : OPA ! 1994 (M. Guyon, J. Escobar, M. Imbard) Original code |
---|
7 | !! 7.0 ! 1997 (A.M. Treguier) SHMEM additions |
---|
8 | !! 8.0 ! 1998 (M. Imbard, J. Escobar, L. Colombet ) SHMEM and MPI |
---|
9 | !! ! 1998 (J.M. Molines) Open boundary conditions |
---|
10 | !! NEMO 1.0 ! 2003 (J.-M. Molines, G. Madec) F90, free form |
---|
11 | !! ! 2003 (J.M. Molines) add mpp_ini_north(_3d,_2d) |
---|
12 | !! - ! 2004 (R. Bourdalle Badie) isend option in mpi |
---|
13 | !! ! 2004 (J.M. Molines) minloc, maxloc |
---|
14 | !! - ! 2005 (G. Madec, S. Masson) npolj=5,6 F-point & ice cases |
---|
15 | !! - ! 2005 (R. Redler) Replacement of MPI_COMM_WORLD except for MPI_Abort |
---|
16 | !! - ! 2005 (R. Benshila, G. Madec) add extra halo case |
---|
17 | !! - ! 2008 (R. Benshila) add mpp_ini_ice |
---|
18 | !! 3.2 ! 2009 (R. Benshila) SHMEM suppression, north fold in lbc_nfd |
---|
19 | !! 3.2 ! 2009 (O. Marti) add mpp_ini_znl |
---|
20 | !!---------------------------------------------------------------------- |
---|
21 | #if defined key_mpp_mpi |
---|
22 | !!---------------------------------------------------------------------- |
---|
23 | !! 'key_mpp_mpi' MPI massively parallel processing library |
---|
24 | !!---------------------------------------------------------------------- |
---|
25 | !! mynode : indentify the processor unit |
---|
26 | !! mpp_lnk : interface (defined in lbclnk) for message passing of 2d or 3d arrays (mpp_lnk_2d, mpp_lnk_3d) |
---|
27 | !! mpp_lnk_3d_gather : Message passing manadgement for two 3D arrays |
---|
28 | !! mpp_lnk_e : interface (defined in lbclnk) for message passing of 2d array with extra halo (mpp_lnk_2d_e) |
---|
29 | !! mpprecv : |
---|
30 | !! mppsend : SUBROUTINE mpp_ini_znl |
---|
31 | !! mppscatter : |
---|
32 | !! mppgather : |
---|
33 | !! mpp_min : generic interface for mppmin_int , mppmin_a_int , mppmin_real, mppmin_a_real |
---|
34 | !! mpp_max : generic interface for mppmax_int , mppmax_a_int , mppmax_real, mppmax_a_real |
---|
35 | !! mpp_sum : generic interface for mppsum_int , mppsum_a_int , mppsum_real, mppsum_a_real |
---|
36 | !! mpp_minloc : |
---|
37 | !! mpp_maxloc : |
---|
38 | !! mppsync : |
---|
39 | !! mppstop : |
---|
40 | !! mppobc : variant of mpp_lnk for open boundary condition |
---|
41 | !! mpp_ini_north : initialisation of north fold |
---|
42 | !! mpp_lbc_north : north fold processors gathering |
---|
43 | !! mpp_lbc_north_e : variant of mpp_lbc_north for extra outer halo |
---|
44 | !!---------------------------------------------------------------------- |
---|
45 | !! History : |
---|
46 | !! ! 94 (M. Guyon, J. Escobar, M. Imbard) Original code |
---|
47 | !! ! 97 (A.M. Treguier) SHMEM additions |
---|
48 | !! ! 98 (M. Imbard, J. Escobar, L. Colombet ) SHMEM and MPI |
---|
49 | !! 9.0 ! 03 (J.-M. Molines, G. Madec) F90, free form |
---|
50 | !! ! 04 (R. Bourdalle Badie) isend option in mpi |
---|
51 | !! ! 05 (G. Madec, S. Masson) npolj=5,6 F-point & ice cases |
---|
52 | !! ! 05 (R. Redler) Replacement of MPI_COMM_WORLD except for MPI_Abort |
---|
53 | !! ! 09 (R. Benshila) SHMEM suppression, north fold in lbc_nfd |
---|
54 | !!---------------------------------------------------------------------- |
---|
55 | !! OPA 9.0 , LOCEAN-IPSL (2005) |
---|
56 | !! $Id$ |
---|
57 | !! This software is governed by the CeCILL licence see modipsl/doc/NEMO_CeCILL.txt |
---|
58 | !!--------------------------------------------------------------------- |
---|
59 | !! * Modules used |
---|
60 | USE dom_oce ! ocean space and time domain |
---|
61 | USE in_out_manager ! I/O manager |
---|
62 | USE lbcnfd ! north fold treatment |
---|
63 | |
---|
64 | IMPLICIT NONE |
---|
65 | PRIVATE |
---|
66 | |
---|
67 | PUBLIC mynode, mppstop, mppsync, mpp_comm_free |
---|
68 | PUBLIC mpp_ini_north, mpp_lbc_north, mpp_lbc_north_e |
---|
69 | PUBLIC mpp_min, mpp_max, mpp_sum, mpp_minloc, mpp_maxloc |
---|
70 | PUBLIC mpp_lnk_3d, mpp_lnk_3d_gather, mpp_lnk_2d, mpp_lnk_2d_e |
---|
71 | PUBLIC mpprecv, mppsend, mppscatter, mppgather |
---|
72 | PUBLIC mppobc, mpp_ini_ice, mpp_ini_znl |
---|
73 | #if defined key_oasis3 || defined key_oasis4 |
---|
74 | PUBLIC mppsize, mpprank |
---|
75 | #endif |
---|
76 | |
---|
77 | !! * Interfaces |
---|
78 | !! define generic interface for these routine as they are called sometimes |
---|
79 | !! with scalar arguments instead of array arguments, which causes problems |
---|
80 | !! for the compilation on AIX system as well as NEC and SGI. Ok on COMPACQ |
---|
81 | INTERFACE mpp_min |
---|
82 | MODULE PROCEDURE mppmin_a_int, mppmin_int, mppmin_a_real, mppmin_real |
---|
83 | END INTERFACE |
---|
84 | INTERFACE mpp_max |
---|
85 | MODULE PROCEDURE mppmax_a_int, mppmax_int, mppmax_a_real, mppmax_real |
---|
86 | END INTERFACE |
---|
87 | INTERFACE mpp_sum |
---|
88 | MODULE PROCEDURE mppsum_a_int, mppsum_int, mppsum_a_real, mppsum_real |
---|
89 | END INTERFACE |
---|
90 | INTERFACE mpp_lbc_north |
---|
91 | MODULE PROCEDURE mpp_lbc_north_3d, mpp_lbc_north_2d |
---|
92 | END INTERFACE |
---|
93 | INTERFACE mpp_minloc |
---|
94 | MODULE PROCEDURE mpp_minloc2d ,mpp_minloc3d |
---|
95 | END INTERFACE |
---|
96 | INTERFACE mpp_maxloc |
---|
97 | MODULE PROCEDURE mpp_maxloc2d ,mpp_maxloc3d |
---|
98 | END INTERFACE |
---|
99 | |
---|
100 | |
---|
101 | !! ========================= !! |
---|
102 | !! MPI variable definition !! |
---|
103 | !! ========================= !! |
---|
104 | !$AGRIF_DO_NOT_TREAT |
---|
105 | # include <mpif.h> |
---|
106 | !$AGRIF_END_DO_NOT_TREAT |
---|
107 | |
---|
108 | LOGICAL, PUBLIC, PARAMETER :: lk_mpp = .TRUE. !: mpp flag |
---|
109 | |
---|
110 | INTEGER, PARAMETER :: nprocmax = 2**10 ! maximun dimension (required to be a power of 2) |
---|
111 | |
---|
112 | INTEGER :: mppsize ! number of process |
---|
113 | INTEGER :: mpprank ! process number [ 0 - size-1 ] |
---|
114 | !$AGRIF_DO_NOT_TREAT |
---|
115 | INTEGER, PUBLIC :: mpi_comm_opa ! opa local communicator |
---|
116 | !$AGRIF_END_DO_NOT_TREAT |
---|
117 | |
---|
118 | ! variables used in case of sea-ice |
---|
119 | INTEGER, PUBLIC :: ncomm_ice !: communicator made by the processors with sea-ice |
---|
120 | INTEGER :: ngrp_ice ! group ID for the ice processors (for rheology) |
---|
121 | INTEGER :: ndim_rank_ice ! number of 'ice' processors |
---|
122 | INTEGER :: n_ice_root ! number (in the comm_ice) of proc 0 in the ice comm |
---|
123 | INTEGER, DIMENSION(:), ALLOCATABLE :: nrank_ice ! dimension ndim_rank_ice |
---|
124 | |
---|
125 | ! variables used for zonal integration |
---|
126 | INTEGER, PUBLIC :: ncomm_znl !: communicator made by the processors on the same zonal average |
---|
127 | LOGICAL, PUBLIC :: l_znl_root ! True on the 'left'most processor on the same row |
---|
128 | INTEGER :: ngrp_znl ! group ID for the znl processors |
---|
129 | INTEGER :: ndim_rank_znl ! number of processors on the same zonal average |
---|
130 | INTEGER, DIMENSION(:), ALLOCATABLE :: nrank_znl ! dimension ndim_rank_znl, number of the procs into the same znl domain |
---|
131 | |
---|
132 | ! North fold condition in mpp_mpi with jpni > 1 |
---|
133 | INTEGER :: ngrp_world ! group ID for the world processors |
---|
134 | INTEGER :: ngrp_opa ! group ID for the opa processors |
---|
135 | INTEGER :: ngrp_north ! group ID for the northern processors (to be fold) |
---|
136 | INTEGER :: ncomm_north ! communicator made by the processors belonging to ngrp_north |
---|
137 | INTEGER :: ndim_rank_north ! number of 'sea' processor in the northern line (can be /= jpni !) |
---|
138 | INTEGER :: njmppmax ! value of njmpp for the processors of the northern line |
---|
139 | INTEGER :: north_root ! number (in the comm_opa) of proc 0 in the northern comm |
---|
140 | INTEGER, DIMENSION(:), ALLOCATABLE :: nrank_north ! dimension ndim_rank_north |
---|
141 | |
---|
142 | ! Type of send : standard, buffered, immediate |
---|
143 | CHARACTER(len=1) :: cn_mpi_send = 'S' ! type od mpi send/recieve (S=standard, B=bsend, I=isend) |
---|
144 | LOGICAL :: l_isend = .FALSE. ! isend use indicator (T if cn_mpi_send='I') |
---|
145 | INTEGER :: nn_buffer = 0 ! size of the buffer in case of mpi_bsend |
---|
146 | |
---|
147 | REAL(wp), ALLOCATABLE, DIMENSION(:) :: tampon ! buffer in case of bsend |
---|
148 | |
---|
149 | ! message passing arrays |
---|
150 | REAL(wp), DIMENSION(jpi,jprecj,jpk,2,2) :: t4ns, t4sn ! 2 x 3d for north-south & south-north |
---|
151 | REAL(wp), DIMENSION(jpj,jpreci,jpk,2,2) :: t4ew, t4we ! 2 x 3d for east-west & west-east |
---|
152 | REAL(wp), DIMENSION(jpi,jprecj,jpk,2,2) :: t4p1, t4p2 ! 2 x 3d for north fold |
---|
153 | REAL(wp), DIMENSION(jpi,jprecj,jpk,2) :: t3ns, t3sn ! 3d for north-south & south-north |
---|
154 | REAL(wp), DIMENSION(jpj,jpreci,jpk,2) :: t3ew, t3we ! 3d for east-west & west-east |
---|
155 | REAL(wp), DIMENSION(jpi,jprecj,jpk,2) :: t3p1, t3p2 ! 3d for north fold |
---|
156 | REAL(wp), DIMENSION(jpi,jprecj,2) :: t2ns, t2sn ! 2d for north-south & south-north |
---|
157 | REAL(wp), DIMENSION(jpj,jpreci,2) :: t2ew, t2we ! 2d for east-west & west-east |
---|
158 | REAL(wp), DIMENSION(jpi,jprecj,2) :: t2p1, t2p2 ! 2d for north fold |
---|
159 | REAL(wp), DIMENSION(1-jpr2di:jpi+jpr2di,jprecj+jpr2dj,2) :: tr2ns, tr2sn ! 2d for north-south & south-north + extra outer halo |
---|
160 | REAL(wp), DIMENSION(1-jpr2dj:jpj+jpr2dj,jpreci+jpr2di,2) :: tr2ew, tr2we ! 2d for east-west & west-east + extra outer halo |
---|
161 | |
---|
162 | ! North fold arrays used to minimise the use of allgather operations. Set in opa_northcomms so need to be public |
---|
163 | INTEGER, PUBLIC, PARAMETER :: jpmaxngh = 8 ! Assumed maximum number of active neighbours |
---|
164 | INTEGER, PUBLIC, DIMENSION (jpmaxngh,4) :: isendto |
---|
165 | INTEGER, PUBLIC, DIMENSION (4) :: nsndto |
---|
166 | LOGICAL, PUBLIC :: lnorth_nogather = .FALSE. |
---|
167 | INTEGER, PUBLIC :: ityp |
---|
168 | !!---------------------------------------------------------------------- |
---|
169 | !! NEMO/OPA 3.2 , LOCEAN-IPSL (2009) |
---|
170 | !! $Id$ |
---|
171 | !! Software governed by the CeCILL licence (modipsl/doc/NEMO_CeCILL.txt) |
---|
172 | !!---------------------------------------------------------------------- |
---|
173 | |
---|
174 | CONTAINS |
---|
175 | |
---|
176 | FUNCTION mynode(ldtxt, localComm) |
---|
177 | !!---------------------------------------------------------------------- |
---|
178 | !! *** routine mynode *** |
---|
179 | !! |
---|
180 | !! ** Purpose : Find processor unit |
---|
181 | !! |
---|
182 | !!---------------------------------------------------------------------- |
---|
183 | CHARACTER(len=*),DIMENSION(:), INTENT( out) :: ldtxt |
---|
184 | INTEGER, OPTIONAL , INTENT(in ) :: localComm |
---|
185 | INTEGER :: mynode, ierr, code |
---|
186 | LOGICAL :: mpi_was_called |
---|
187 | |
---|
188 | NAMELIST/nammpp/ cn_mpi_send, nn_buffer |
---|
189 | !!---------------------------------------------------------------------- |
---|
190 | ! |
---|
191 | WRITE(ldtxt(1),*) |
---|
192 | WRITE(ldtxt(2),*) 'mynode : mpi initialisation' |
---|
193 | WRITE(ldtxt(3),*) '~~~~~~ ' |
---|
194 | ! |
---|
195 | REWIND( numnam ) ! Namelist namrun : parameters of the run |
---|
196 | READ ( numnam, nammpp ) |
---|
197 | ! ! control print |
---|
198 | WRITE(ldtxt(4),*) ' Namelist nammpp' |
---|
199 | WRITE(ldtxt(5),*) ' mpi send type cn_mpi_send = ', cn_mpi_send |
---|
200 | WRITE(ldtxt(6),*) ' size in bytes of exported buffer nn_buffer = ', nn_buffer |
---|
201 | |
---|
202 | CALL mpi_initialized ( mpi_was_called, code ) |
---|
203 | IF( code /= MPI_SUCCESS ) THEN |
---|
204 | WRITE(*, cform_err) |
---|
205 | WRITE(*, *) 'lib_mpp: Error in routine mpi_initialized' |
---|
206 | CALL mpi_abort( mpi_comm_world, code, ierr ) |
---|
207 | ENDIF |
---|
208 | |
---|
209 | IF( mpi_was_called ) THEN |
---|
210 | ! |
---|
211 | SELECT CASE ( cn_mpi_send ) |
---|
212 | CASE ( 'S' ) ! Standard mpi send (blocking) |
---|
213 | WRITE(ldtxt(7),*) ' Standard blocking mpi send (send)' |
---|
214 | CASE ( 'B' ) ! Buffer mpi send (blocking) |
---|
215 | WRITE(ldtxt(7),*) ' Buffer blocking mpi send (bsend)' |
---|
216 | CALL mpi_init_opa( ierr ) |
---|
217 | CASE ( 'I' ) ! Immediate mpi send (non-blocking send) |
---|
218 | WRITE(ldtxt(7),*) ' Immediate non-blocking send (isend)' |
---|
219 | l_isend = .TRUE. |
---|
220 | CASE DEFAULT |
---|
221 | WRITE(ldtxt(7),cform_err) |
---|
222 | WRITE(ldtxt(8),*) ' bad value for cn_mpi_send = ', cn_mpi_send |
---|
223 | nstop = nstop + 1 |
---|
224 | END SELECT |
---|
225 | ELSE IF ( PRESENT(localComm) .and. .not. mpi_was_called ) THEN |
---|
226 | WRITE(ldtxt(7),*) ' lib_mpp: You cannot provide a local communicator ' |
---|
227 | WRITE(ldtxt(8),*) ' without calling MPI_Init before ! ' |
---|
228 | nstop = nstop + 1 |
---|
229 | ELSE |
---|
230 | SELECT CASE ( cn_mpi_send ) |
---|
231 | CASE ( 'S' ) ! Standard mpi send (blocking) |
---|
232 | WRITE(ldtxt(7),*) ' Standard blocking mpi send (send)' |
---|
233 | CALL mpi_init( ierr ) |
---|
234 | CASE ( 'B' ) ! Buffer mpi send (blocking) |
---|
235 | WRITE(ldtxt(7),*) ' Buffer blocking mpi send (bsend)' |
---|
236 | CALL mpi_init_opa( ierr ) |
---|
237 | CASE ( 'I' ) ! Immediate mpi send (non-blocking send) |
---|
238 | WRITE(ldtxt(7),*) ' Immediate non-blocking send (isend)' |
---|
239 | l_isend = .TRUE. |
---|
240 | CALL mpi_init( ierr ) |
---|
241 | CASE DEFAULT |
---|
242 | WRITE(ldtxt(7),cform_err) |
---|
243 | WRITE(ldtxt(8),*) ' bad value for cn_mpi_send = ', cn_mpi_send |
---|
244 | nstop = nstop + 1 |
---|
245 | END SELECT |
---|
246 | ! |
---|
247 | ENDIF |
---|
248 | |
---|
249 | IF( PRESENT(localComm) ) THEN |
---|
250 | IF( Agrif_Root() ) THEN |
---|
251 | mpi_comm_opa = localComm |
---|
252 | ENDIF |
---|
253 | ELSE |
---|
254 | CALL mpi_comm_dup( mpi_comm_world, mpi_comm_opa, code) |
---|
255 | IF( code /= MPI_SUCCESS ) THEN |
---|
256 | WRITE(*, cform_err) |
---|
257 | WRITE(*, *) ' lib_mpp: Error in routine mpi_comm_dup' |
---|
258 | CALL mpi_abort( mpi_comm_world, code, ierr ) |
---|
259 | ENDIF |
---|
260 | ENDIF |
---|
261 | |
---|
262 | CALL mpi_comm_rank( mpi_comm_opa, mpprank, ierr ) |
---|
263 | CALL mpi_comm_size( mpi_comm_opa, mppsize, ierr ) |
---|
264 | mynode = mpprank |
---|
265 | ! |
---|
266 | END FUNCTION mynode |
---|
267 | |
---|
268 | |
---|
269 | SUBROUTINE mpp_lnk_3d( ptab, cd_type, psgn, cd_mpp, pval ) |
---|
270 | !!---------------------------------------------------------------------- |
---|
271 | !! *** routine mpp_lnk_3d *** |
---|
272 | !! |
---|
273 | !! ** Purpose : Message passing manadgement |
---|
274 | !! |
---|
275 | !! ** Method : Use mppsend and mpprecv function for passing mask |
---|
276 | !! between processors following neighboring subdomains. |
---|
277 | !! domain parameters |
---|
278 | !! nlci : first dimension of the local subdomain |
---|
279 | !! nlcj : second dimension of the local subdomain |
---|
280 | !! nbondi : mark for "east-west local boundary" |
---|
281 | !! nbondj : mark for "north-south local boundary" |
---|
282 | !! noea : number for local neighboring processors |
---|
283 | !! nowe : number for local neighboring processors |
---|
284 | !! noso : number for local neighboring processors |
---|
285 | !! nono : number for local neighboring processors |
---|
286 | !! |
---|
287 | !! ** Action : ptab with update value at its periphery |
---|
288 | !! |
---|
289 | !!---------------------------------------------------------------------- |
---|
290 | REAL(wp), DIMENSION(jpi,jpj,jpk), INTENT(inout) :: ptab ! 3D array on which the boundary condition is applied |
---|
291 | CHARACTER(len=1) , INTENT(in ) :: cd_type ! define the nature of ptab array grid-points |
---|
292 | ! ! = T , U , V , F , W points |
---|
293 | REAL(wp) , INTENT(in ) :: psgn ! =-1 the sign change across the north fold boundary |
---|
294 | ! ! = 1. , the sign is kept |
---|
295 | CHARACTER(len=3), OPTIONAL , INTENT(in ) :: cd_mpp ! fill the overlap area only |
---|
296 | REAL(wp) , OPTIONAL , INTENT(in ) :: pval ! background value (used at closed boundaries) |
---|
297 | !! |
---|
298 | INTEGER :: ji, jj, jk, jl ! dummy loop indices |
---|
299 | INTEGER :: imigr, iihom, ijhom ! temporary integers |
---|
300 | INTEGER :: ml_req1, ml_req2, ml_err ! for key_mpi_isend |
---|
301 | REAL(wp) :: zland |
---|
302 | INTEGER, DIMENSION(MPI_STATUS_SIZE) :: ml_stat ! for key_mpi_isend |
---|
303 | !!---------------------------------------------------------------------- |
---|
304 | |
---|
305 | IF( PRESENT( pval ) ) THEN ; zland = pval ! set land value |
---|
306 | ELSE ; zland = 0.e0 ! zero by default |
---|
307 | ENDIF |
---|
308 | |
---|
309 | ! 1. standard boundary treatment |
---|
310 | ! ------------------------------ |
---|
311 | IF( PRESENT( cd_mpp ) ) THEN ! only fill added line/raw with existing values |
---|
312 | ! |
---|
313 | ! WARNING ptab is defined only between nld and nle |
---|
314 | DO jk = 1, jpk |
---|
315 | DO jj = nlcj+1, jpj ! added line(s) (inner only) |
---|
316 | ptab(nldi :nlei , jj ,jk) = ptab(nldi:nlei, nlej,jk) |
---|
317 | ptab(1 :nldi-1, jj ,jk) = ptab(nldi , nlej,jk) |
---|
318 | ptab(nlei+1:nlci , jj ,jk) = ptab( nlei, nlej,jk) |
---|
319 | END DO |
---|
320 | DO ji = nlci+1, jpi ! added column(s) (full) |
---|
321 | ptab(ji ,nldj :nlej ,jk) = ptab( nlei,nldj:nlej,jk) |
---|
322 | ptab(ji ,1 :nldj-1,jk) = ptab( nlei,nldj ,jk) |
---|
323 | ptab(ji ,nlej+1:jpj ,jk) = ptab( nlei, nlej,jk) |
---|
324 | END DO |
---|
325 | END DO |
---|
326 | ! |
---|
327 | ELSE ! standard close or cyclic treatment |
---|
328 | ! |
---|
329 | ! ! East-West boundaries |
---|
330 | ! !* Cyclic east-west |
---|
331 | IF( nbondi == 2 .AND. (nperio == 1 .OR. nperio == 4 .OR. nperio == 6) ) THEN |
---|
332 | ptab( 1 ,:,:) = ptab(jpim1,:,:) |
---|
333 | ptab(jpi,:,:) = ptab( 2 ,:,:) |
---|
334 | ELSE !* closed |
---|
335 | IF( .NOT. cd_type == 'F' ) ptab( 1 :jpreci,:,:) = zland ! south except F-point |
---|
336 | ptab(nlci-jpreci+1:jpi ,:,:) = zland ! north |
---|
337 | ENDIF |
---|
338 | ! ! North-South boundaries (always closed) |
---|
339 | IF( .NOT. cd_type == 'F' ) ptab(:, 1 :jprecj,:) = zland ! south except F-point |
---|
340 | ptab(:,nlcj-jprecj+1:jpj ,:) = zland ! north |
---|
341 | ! |
---|
342 | ENDIF |
---|
343 | |
---|
344 | ! 2. East and west directions exchange |
---|
345 | ! ------------------------------------ |
---|
346 | ! we play with the neigbours AND the row number because of the periodicity |
---|
347 | ! |
---|
348 | SELECT CASE ( nbondi ) ! Read Dirichlet lateral conditions |
---|
349 | CASE ( -1, 0, 1 ) ! all exept 2 (i.e. close case) |
---|
350 | iihom = nlci-nreci |
---|
351 | DO jl = 1, jpreci |
---|
352 | t3ew(:,jl,:,1) = ptab(jpreci+jl,:,:) |
---|
353 | t3we(:,jl,:,1) = ptab(iihom +jl,:,:) |
---|
354 | END DO |
---|
355 | END SELECT |
---|
356 | ! |
---|
357 | ! ! Migrations |
---|
358 | imigr = jpreci * jpj * jpk |
---|
359 | ! |
---|
360 | SELECT CASE ( nbondi ) |
---|
361 | CASE ( -1 ) |
---|
362 | CALL mppsend( 2, t3we(1,1,1,1), imigr, noea, ml_req1 ) |
---|
363 | CALL mpprecv( 1, t3ew(1,1,1,2), imigr, noea ) |
---|
364 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
365 | CASE ( 0 ) |
---|
366 | CALL mppsend( 1, t3ew(1,1,1,1), imigr, nowe, ml_req1 ) |
---|
367 | CALL mppsend( 2, t3we(1,1,1,1), imigr, noea, ml_req2 ) |
---|
368 | CALL mpprecv( 1, t3ew(1,1,1,2), imigr, noea ) |
---|
369 | CALL mpprecv( 2, t3we(1,1,1,2), imigr, nowe ) |
---|
370 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
371 | IF(l_isend) CALL mpi_wait(ml_req2, ml_stat, ml_err) |
---|
372 | CASE ( 1 ) |
---|
373 | CALL mppsend( 1, t3ew(1,1,1,1), imigr, nowe, ml_req1 ) |
---|
374 | CALL mpprecv( 2, t3we(1,1,1,2), imigr, nowe ) |
---|
375 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
376 | END SELECT |
---|
377 | ! |
---|
378 | ! ! Write Dirichlet lateral conditions |
---|
379 | iihom = nlci-jpreci |
---|
380 | ! |
---|
381 | SELECT CASE ( nbondi ) |
---|
382 | CASE ( -1 ) |
---|
383 | DO jl = 1, jpreci |
---|
384 | ptab(iihom+jl,:,:) = t3ew(:,jl,:,2) |
---|
385 | END DO |
---|
386 | CASE ( 0 ) |
---|
387 | DO jl = 1, jpreci |
---|
388 | ptab(jl ,:,:) = t3we(:,jl,:,2) |
---|
389 | ptab(iihom+jl,:,:) = t3ew(:,jl,:,2) |
---|
390 | END DO |
---|
391 | CASE ( 1 ) |
---|
392 | DO jl = 1, jpreci |
---|
393 | ptab(jl ,:,:) = t3we(:,jl,:,2) |
---|
394 | END DO |
---|
395 | END SELECT |
---|
396 | |
---|
397 | |
---|
398 | ! 3. North and south directions |
---|
399 | ! ----------------------------- |
---|
400 | ! always closed : we play only with the neigbours |
---|
401 | ! |
---|
402 | IF( nbondj /= 2 ) THEN ! Read Dirichlet lateral conditions |
---|
403 | ijhom = nlcj-nrecj |
---|
404 | DO jl = 1, jprecj |
---|
405 | t3sn(:,jl,:,1) = ptab(:,ijhom +jl,:) |
---|
406 | t3ns(:,jl,:,1) = ptab(:,jprecj+jl,:) |
---|
407 | END DO |
---|
408 | ENDIF |
---|
409 | ! |
---|
410 | ! ! Migrations |
---|
411 | imigr = jprecj * jpi * jpk |
---|
412 | ! |
---|
413 | SELECT CASE ( nbondj ) |
---|
414 | CASE ( -1 ) |
---|
415 | CALL mppsend( 4, t3sn(1,1,1,1), imigr, nono, ml_req1 ) |
---|
416 | CALL mpprecv( 3, t3ns(1,1,1,2), imigr, nono ) |
---|
417 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
418 | CASE ( 0 ) |
---|
419 | CALL mppsend( 3, t3ns(1,1,1,1), imigr, noso, ml_req1 ) |
---|
420 | CALL mppsend( 4, t3sn(1,1,1,1), imigr, nono, ml_req2 ) |
---|
421 | CALL mpprecv( 3, t3ns(1,1,1,2), imigr, nono ) |
---|
422 | CALL mpprecv( 4, t3sn(1,1,1,2), imigr, noso ) |
---|
423 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
424 | IF(l_isend) CALL mpi_wait(ml_req2, ml_stat, ml_err) |
---|
425 | CASE ( 1 ) |
---|
426 | CALL mppsend( 3, t3ns(1,1,1,1), imigr, noso, ml_req1 ) |
---|
427 | CALL mpprecv( 4, t3sn(1,1,1,2), imigr, noso ) |
---|
428 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
429 | END SELECT |
---|
430 | ! |
---|
431 | ! ! Write Dirichlet lateral conditions |
---|
432 | ijhom = nlcj-jprecj |
---|
433 | ! |
---|
434 | SELECT CASE ( nbondj ) |
---|
435 | CASE ( -1 ) |
---|
436 | DO jl = 1, jprecj |
---|
437 | ptab(:,ijhom+jl,:) = t3ns(:,jl,:,2) |
---|
438 | END DO |
---|
439 | CASE ( 0 ) |
---|
440 | DO jl = 1, jprecj |
---|
441 | ptab(:,jl ,:) = t3sn(:,jl,:,2) |
---|
442 | ptab(:,ijhom+jl,:) = t3ns(:,jl,:,2) |
---|
443 | END DO |
---|
444 | CASE ( 1 ) |
---|
445 | DO jl = 1, jprecj |
---|
446 | ptab(:,jl,:) = t3sn(:,jl,:,2) |
---|
447 | END DO |
---|
448 | END SELECT |
---|
449 | |
---|
450 | |
---|
451 | ! 4. north fold treatment |
---|
452 | ! ----------------------- |
---|
453 | ! |
---|
454 | IF( npolj /= 0 .AND. .NOT. PRESENT(cd_mpp) ) THEN |
---|
455 | ! |
---|
456 | SELECT CASE ( jpni ) |
---|
457 | CASE ( 1 ) ; CALL lbc_nfd ( ptab, cd_type, psgn ) ! only 1 northern proc, no mpp |
---|
458 | CASE DEFAULT ; CALL mpp_lbc_north( ptab, cd_type, psgn ) ! for all northern procs. |
---|
459 | END SELECT |
---|
460 | ! |
---|
461 | ENDIF |
---|
462 | ! |
---|
463 | END SUBROUTINE mpp_lnk_3d |
---|
464 | |
---|
465 | |
---|
466 | SUBROUTINE mpp_lnk_2d( pt2d, cd_type, psgn, cd_mpp, pval ) |
---|
467 | !!---------------------------------------------------------------------- |
---|
468 | !! *** routine mpp_lnk_2d *** |
---|
469 | !! |
---|
470 | !! ** Purpose : Message passing manadgement for 2d array |
---|
471 | !! |
---|
472 | !! ** Method : Use mppsend and mpprecv function for passing mask |
---|
473 | !! between processors following neighboring subdomains. |
---|
474 | !! domain parameters |
---|
475 | !! nlci : first dimension of the local subdomain |
---|
476 | !! nlcj : second dimension of the local subdomain |
---|
477 | !! nbondi : mark for "east-west local boundary" |
---|
478 | !! nbondj : mark for "north-south local boundary" |
---|
479 | !! noea : number for local neighboring processors |
---|
480 | !! nowe : number for local neighboring processors |
---|
481 | !! noso : number for local neighboring processors |
---|
482 | !! nono : number for local neighboring processors |
---|
483 | !! |
---|
484 | !!---------------------------------------------------------------------- |
---|
485 | REAL(wp), DIMENSION(jpi,jpj), INTENT(inout) :: pt2d ! 2D array on which the boundary condition is applied |
---|
486 | CHARACTER(len=1) , INTENT(in ) :: cd_type ! define the nature of ptab array grid-points |
---|
487 | ! ! = T , U , V , F , W and I points |
---|
488 | REAL(wp) , INTENT(in ) :: psgn ! =-1 the sign change across the north fold boundary |
---|
489 | ! ! = 1. , the sign is kept |
---|
490 | CHARACTER(len=3), OPTIONAL , INTENT(in ) :: cd_mpp ! fill the overlap area only |
---|
491 | REAL(wp) , OPTIONAL , INTENT(in ) :: pval ! background value (used at closed boundaries) |
---|
492 | !! |
---|
493 | INTEGER :: ji, jj, jl ! dummy loop indices |
---|
494 | INTEGER :: imigr, iihom, ijhom ! temporary integers |
---|
495 | INTEGER :: ml_req1, ml_req2, ml_err ! for key_mpi_isend |
---|
496 | REAL(wp) :: zland |
---|
497 | INTEGER, DIMENSION(MPI_STATUS_SIZE) :: ml_stat ! for key_mpi_isend |
---|
498 | !!---------------------------------------------------------------------- |
---|
499 | |
---|
500 | IF( PRESENT( pval ) ) THEN ; zland = pval ! set land value |
---|
501 | ELSE ; zland = 0.e0 ! zero by default |
---|
502 | ENDIF |
---|
503 | |
---|
504 | ! 1. standard boundary treatment |
---|
505 | ! ------------------------------ |
---|
506 | ! |
---|
507 | IF( PRESENT( cd_mpp ) ) THEN ! only fill added line/raw with existing values |
---|
508 | ! |
---|
509 | ! WARNING pt2d is defined only between nld and nle |
---|
510 | DO jj = nlcj+1, jpj ! added line(s) (inner only) |
---|
511 | pt2d(nldi :nlei , jj ) = pt2d(nldi:nlei, nlej) |
---|
512 | pt2d(1 :nldi-1, jj ) = pt2d(nldi , nlej) |
---|
513 | pt2d(nlei+1:nlci , jj ) = pt2d( nlei, nlej) |
---|
514 | END DO |
---|
515 | DO ji = nlci+1, jpi ! added column(s) (full) |
---|
516 | pt2d(ji ,nldj :nlej ) = pt2d( nlei,nldj:nlej) |
---|
517 | pt2d(ji ,1 :nldj-1) = pt2d( nlei,nldj ) |
---|
518 | pt2d(ji ,nlej+1:jpj ) = pt2d( nlei, nlej) |
---|
519 | END DO |
---|
520 | ! |
---|
521 | ELSE ! standard close or cyclic treatment |
---|
522 | ! |
---|
523 | ! ! East-West boundaries |
---|
524 | IF( nbondi == 2 .AND. & ! Cyclic east-west |
---|
525 | & (nperio == 1 .OR. nperio == 4 .OR. nperio == 6) ) THEN |
---|
526 | pt2d( 1 ,:) = pt2d(jpim1,:) ! west |
---|
527 | pt2d(jpi,:) = pt2d( 2 ,:) ! east |
---|
528 | ELSE ! closed |
---|
529 | IF( .NOT. cd_type == 'F' ) pt2d( 1 :jpreci,:) = zland ! south except F-point |
---|
530 | pt2d(nlci-jpreci+1:jpi ,:) = zland ! north |
---|
531 | ENDIF |
---|
532 | ! ! North-South boundaries (always closed) |
---|
533 | IF( .NOT. cd_type == 'F' ) pt2d(:, 1 :jprecj) = zland !south except F-point |
---|
534 | pt2d(:,nlcj-jprecj+1:jpj ) = zland ! north |
---|
535 | ! |
---|
536 | ENDIF |
---|
537 | |
---|
538 | ! 2. East and west directions exchange |
---|
539 | ! ------------------------------------ |
---|
540 | ! we play with the neigbours AND the row number because of the periodicity |
---|
541 | ! |
---|
542 | SELECT CASE ( nbondi ) ! Read Dirichlet lateral conditions |
---|
543 | CASE ( -1, 0, 1 ) ! all exept 2 (i.e. close case) |
---|
544 | iihom = nlci-nreci |
---|
545 | DO jl = 1, jpreci |
---|
546 | t2ew(:,jl,1) = pt2d(jpreci+jl,:) |
---|
547 | t2we(:,jl,1) = pt2d(iihom +jl,:) |
---|
548 | END DO |
---|
549 | END SELECT |
---|
550 | ! |
---|
551 | ! ! Migrations |
---|
552 | imigr = jpreci * jpj |
---|
553 | ! |
---|
554 | SELECT CASE ( nbondi ) |
---|
555 | CASE ( -1 ) |
---|
556 | CALL mppsend( 2, t2we(1,1,1), imigr, noea, ml_req1 ) |
---|
557 | CALL mpprecv( 1, t2ew(1,1,2), imigr, noea ) |
---|
558 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
559 | CASE ( 0 ) |
---|
560 | CALL mppsend( 1, t2ew(1,1,1), imigr, nowe, ml_req1 ) |
---|
561 | CALL mppsend( 2, t2we(1,1,1), imigr, noea, ml_req2 ) |
---|
562 | CALL mpprecv( 1, t2ew(1,1,2), imigr, noea ) |
---|
563 | CALL mpprecv( 2, t2we(1,1,2), imigr, nowe ) |
---|
564 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
565 | IF(l_isend) CALL mpi_wait(ml_req2,ml_stat,ml_err) |
---|
566 | CASE ( 1 ) |
---|
567 | CALL mppsend( 1, t2ew(1,1,1), imigr, nowe, ml_req1 ) |
---|
568 | CALL mpprecv( 2, t2we(1,1,2), imigr, nowe ) |
---|
569 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
570 | END SELECT |
---|
571 | ! |
---|
572 | ! ! Write Dirichlet lateral conditions |
---|
573 | iihom = nlci - jpreci |
---|
574 | ! |
---|
575 | SELECT CASE ( nbondi ) |
---|
576 | CASE ( -1 ) |
---|
577 | DO jl = 1, jpreci |
---|
578 | pt2d(iihom+jl,:) = t2ew(:,jl,2) |
---|
579 | END DO |
---|
580 | CASE ( 0 ) |
---|
581 | DO jl = 1, jpreci |
---|
582 | pt2d(jl ,:) = t2we(:,jl,2) |
---|
583 | pt2d(iihom+jl,:) = t2ew(:,jl,2) |
---|
584 | END DO |
---|
585 | CASE ( 1 ) |
---|
586 | DO jl = 1, jpreci |
---|
587 | pt2d(jl ,:) = t2we(:,jl,2) |
---|
588 | END DO |
---|
589 | END SELECT |
---|
590 | |
---|
591 | |
---|
592 | ! 3. North and south directions |
---|
593 | ! ----------------------------- |
---|
594 | ! always closed : we play only with the neigbours |
---|
595 | ! |
---|
596 | IF( nbondj /= 2 ) THEN ! Read Dirichlet lateral conditions |
---|
597 | ijhom = nlcj-nrecj |
---|
598 | DO jl = 1, jprecj |
---|
599 | t2sn(:,jl,1) = pt2d(:,ijhom +jl) |
---|
600 | t2ns(:,jl,1) = pt2d(:,jprecj+jl) |
---|
601 | END DO |
---|
602 | ENDIF |
---|
603 | ! |
---|
604 | ! ! Migrations |
---|
605 | imigr = jprecj * jpi |
---|
606 | ! |
---|
607 | SELECT CASE ( nbondj ) |
---|
608 | CASE ( -1 ) |
---|
609 | CALL mppsend( 4, t2sn(1,1,1), imigr, nono, ml_req1 ) |
---|
610 | CALL mpprecv( 3, t2ns(1,1,2), imigr, nono ) |
---|
611 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
612 | CASE ( 0 ) |
---|
613 | CALL mppsend( 3, t2ns(1,1,1), imigr, noso, ml_req1 ) |
---|
614 | CALL mppsend( 4, t2sn(1,1,1), imigr, nono, ml_req2 ) |
---|
615 | CALL mpprecv( 3, t2ns(1,1,2), imigr, nono ) |
---|
616 | CALL mpprecv( 4, t2sn(1,1,2), imigr, noso ) |
---|
617 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
618 | IF(l_isend) CALL mpi_wait(ml_req2,ml_stat,ml_err) |
---|
619 | CASE ( 1 ) |
---|
620 | CALL mppsend( 3, t2ns(1,1,1), imigr, noso, ml_req1 ) |
---|
621 | CALL mpprecv( 4, t2sn(1,1,2), imigr, noso ) |
---|
622 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
623 | END SELECT |
---|
624 | ! |
---|
625 | ! ! Write Dirichlet lateral conditions |
---|
626 | ijhom = nlcj - jprecj |
---|
627 | ! |
---|
628 | SELECT CASE ( nbondj ) |
---|
629 | CASE ( -1 ) |
---|
630 | DO jl = 1, jprecj |
---|
631 | pt2d(:,ijhom+jl) = t2ns(:,jl,2) |
---|
632 | END DO |
---|
633 | CASE ( 0 ) |
---|
634 | DO jl = 1, jprecj |
---|
635 | pt2d(:,jl ) = t2sn(:,jl,2) |
---|
636 | pt2d(:,ijhom+jl) = t2ns(:,jl,2) |
---|
637 | END DO |
---|
638 | CASE ( 1 ) |
---|
639 | DO jl = 1, jprecj |
---|
640 | pt2d(:,jl ) = t2sn(:,jl,2) |
---|
641 | END DO |
---|
642 | END SELECT |
---|
643 | |
---|
644 | |
---|
645 | ! 4. north fold treatment |
---|
646 | ! ----------------------- |
---|
647 | ! |
---|
648 | IF( npolj /= 0 .AND. .NOT. PRESENT(cd_mpp) ) THEN |
---|
649 | ! |
---|
650 | SELECT CASE ( jpni ) |
---|
651 | CASE ( 1 ) ; CALL lbc_nfd ( pt2d, cd_type, psgn ) ! only 1 northern proc, no mpp |
---|
652 | CASE DEFAULT ; CALL mpp_lbc_north( pt2d, cd_type, psgn ) ! for all northern procs. |
---|
653 | END SELECT |
---|
654 | ! |
---|
655 | ENDIF |
---|
656 | ! |
---|
657 | END SUBROUTINE mpp_lnk_2d |
---|
658 | |
---|
659 | |
---|
660 | SUBROUTINE mpp_lnk_3d_gather( ptab1, cd_type1, ptab2, cd_type2, psgn ) |
---|
661 | !!---------------------------------------------------------------------- |
---|
662 | !! *** routine mpp_lnk_3d_gather *** |
---|
663 | !! |
---|
664 | !! ** Purpose : Message passing manadgement for two 3D arrays |
---|
665 | !! |
---|
666 | !! ** Method : Use mppsend and mpprecv function for passing mask |
---|
667 | !! between processors following neighboring subdomains. |
---|
668 | !! domain parameters |
---|
669 | !! nlci : first dimension of the local subdomain |
---|
670 | !! nlcj : second dimension of the local subdomain |
---|
671 | !! nbondi : mark for "east-west local boundary" |
---|
672 | !! nbondj : mark for "north-south local boundary" |
---|
673 | !! noea : number for local neighboring processors |
---|
674 | !! nowe : number for local neighboring processors |
---|
675 | !! noso : number for local neighboring processors |
---|
676 | !! nono : number for local neighboring processors |
---|
677 | !! |
---|
678 | !! ** Action : ptab1 and ptab2 with update value at its periphery |
---|
679 | !! |
---|
680 | !!---------------------------------------------------------------------- |
---|
681 | REAL(wp), DIMENSION(jpi,jpj,jpk), INTENT(inout) :: ptab1 ! first and second 3D array on which |
---|
682 | REAL(wp), DIMENSION(jpi,jpj,jpk), INTENT(inout) :: ptab2 ! the boundary condition is applied |
---|
683 | CHARACTER(len=1) , INTENT(in ) :: cd_type1 ! nature of ptab1 and ptab2 arrays |
---|
684 | CHARACTER(len=1) , INTENT(in ) :: cd_type2 ! i.e. grid-points = T , U , V , F or W points |
---|
685 | REAL(wp) , INTENT(in ) :: psgn ! =-1 the sign change across the north fold boundary |
---|
686 | !! ! = 1. , the sign is kept |
---|
687 | INTEGER :: jl ! dummy loop indices |
---|
688 | INTEGER :: imigr, iihom, ijhom ! temporary integers |
---|
689 | INTEGER :: ml_req1, ml_req2, ml_err ! for key_mpi_isend |
---|
690 | INTEGER, DIMENSION(MPI_STATUS_SIZE) :: ml_stat ! for key_mpi_isend |
---|
691 | !!---------------------------------------------------------------------- |
---|
692 | |
---|
693 | ! 1. standard boundary treatment |
---|
694 | ! ------------------------------ |
---|
695 | ! ! East-West boundaries |
---|
696 | ! !* Cyclic east-west |
---|
697 | IF( nbondi == 2 .AND. (nperio == 1 .OR. nperio == 4 .OR. nperio == 6) ) THEN |
---|
698 | ptab1( 1 ,:,:) = ptab1(jpim1,:,:) |
---|
699 | ptab1(jpi,:,:) = ptab1( 2 ,:,:) |
---|
700 | ptab2( 1 ,:,:) = ptab2(jpim1,:,:) |
---|
701 | ptab2(jpi,:,:) = ptab2( 2 ,:,:) |
---|
702 | ELSE !* closed |
---|
703 | IF( .NOT. cd_type1 == 'F' ) ptab1( 1 :jpreci,:,:) = 0.e0 ! south except at F-point |
---|
704 | IF( .NOT. cd_type2 == 'F' ) ptab2( 1 :jpreci,:,:) = 0.e0 |
---|
705 | ptab1(nlci-jpreci+1:jpi ,:,:) = 0.e0 ! north |
---|
706 | ptab2(nlci-jpreci+1:jpi ,:,:) = 0.e0 |
---|
707 | ENDIF |
---|
708 | |
---|
709 | |
---|
710 | ! ! North-South boundaries |
---|
711 | IF( .NOT. cd_type1 == 'F' ) ptab1(:, 1 :jprecj,:) = 0.e0 ! south except at F-point |
---|
712 | IF( .NOT. cd_type2 == 'F' ) ptab2(:, 1 :jprecj,:) = 0.e0 |
---|
713 | ptab1(:,nlcj-jprecj+1:jpj ,:) = 0.e0 ! north |
---|
714 | ptab2(:,nlcj-jprecj+1:jpj ,:) = 0.e0 |
---|
715 | |
---|
716 | |
---|
717 | ! 2. East and west directions exchange |
---|
718 | ! ------------------------------------ |
---|
719 | ! we play with the neigbours AND the row number because of the periodicity |
---|
720 | ! |
---|
721 | SELECT CASE ( nbondi ) ! Read Dirichlet lateral conditions |
---|
722 | CASE ( -1, 0, 1 ) ! all exept 2 (i.e. close case) |
---|
723 | iihom = nlci-nreci |
---|
724 | DO jl = 1, jpreci |
---|
725 | t4ew(:,jl,:,1,1) = ptab1(jpreci+jl,:,:) |
---|
726 | t4we(:,jl,:,1,1) = ptab1(iihom +jl,:,:) |
---|
727 | t4ew(:,jl,:,2,1) = ptab2(jpreci+jl,:,:) |
---|
728 | t4we(:,jl,:,2,1) = ptab2(iihom +jl,:,:) |
---|
729 | END DO |
---|
730 | END SELECT |
---|
731 | ! |
---|
732 | ! ! Migrations |
---|
733 | imigr = jpreci * jpj * jpk *2 |
---|
734 | ! |
---|
735 | SELECT CASE ( nbondi ) |
---|
736 | CASE ( -1 ) |
---|
737 | CALL mppsend( 2, t4we(1,1,1,1,1), imigr, noea, ml_req1 ) |
---|
738 | CALL mpprecv( 1, t4ew(1,1,1,1,2), imigr, noea ) |
---|
739 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
740 | CASE ( 0 ) |
---|
741 | CALL mppsend( 1, t4ew(1,1,1,1,1), imigr, nowe, ml_req1 ) |
---|
742 | CALL mppsend( 2, t4we(1,1,1,1,1), imigr, noea, ml_req2 ) |
---|
743 | CALL mpprecv( 1, t4ew(1,1,1,1,2), imigr, noea ) |
---|
744 | CALL mpprecv( 2, t4we(1,1,1,1,2), imigr, nowe ) |
---|
745 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
746 | IF(l_isend) CALL mpi_wait(ml_req2, ml_stat, ml_err) |
---|
747 | CASE ( 1 ) |
---|
748 | CALL mppsend( 1, t4ew(1,1,1,1,1), imigr, nowe, ml_req1 ) |
---|
749 | CALL mpprecv( 2, t4we(1,1,1,1,2), imigr, nowe ) |
---|
750 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
751 | END SELECT |
---|
752 | ! |
---|
753 | ! ! Write Dirichlet lateral conditions |
---|
754 | iihom = nlci - jpreci |
---|
755 | ! |
---|
756 | SELECT CASE ( nbondi ) |
---|
757 | CASE ( -1 ) |
---|
758 | DO jl = 1, jpreci |
---|
759 | ptab1(iihom+jl,:,:) = t4ew(:,jl,:,1,2) |
---|
760 | ptab2(iihom+jl,:,:) = t4ew(:,jl,:,2,2) |
---|
761 | END DO |
---|
762 | CASE ( 0 ) |
---|
763 | DO jl = 1, jpreci |
---|
764 | ptab1(jl ,:,:) = t4we(:,jl,:,1,2) |
---|
765 | ptab1(iihom+jl,:,:) = t4ew(:,jl,:,1,2) |
---|
766 | ptab2(jl ,:,:) = t4we(:,jl,:,2,2) |
---|
767 | ptab2(iihom+jl,:,:) = t4ew(:,jl,:,2,2) |
---|
768 | END DO |
---|
769 | CASE ( 1 ) |
---|
770 | DO jl = 1, jpreci |
---|
771 | ptab1(jl ,:,:) = t4we(:,jl,:,1,2) |
---|
772 | ptab2(jl ,:,:) = t4we(:,jl,:,2,2) |
---|
773 | END DO |
---|
774 | END SELECT |
---|
775 | |
---|
776 | |
---|
777 | ! 3. North and south directions |
---|
778 | ! ----------------------------- |
---|
779 | ! always closed : we play only with the neigbours |
---|
780 | ! |
---|
781 | IF( nbondj /= 2 ) THEN ! Read Dirichlet lateral conditions |
---|
782 | ijhom = nlcj - nrecj |
---|
783 | DO jl = 1, jprecj |
---|
784 | t4sn(:,jl,:,1,1) = ptab1(:,ijhom +jl,:) |
---|
785 | t4ns(:,jl,:,1,1) = ptab1(:,jprecj+jl,:) |
---|
786 | t4sn(:,jl,:,2,1) = ptab2(:,ijhom +jl,:) |
---|
787 | t4ns(:,jl,:,2,1) = ptab2(:,jprecj+jl,:) |
---|
788 | END DO |
---|
789 | ENDIF |
---|
790 | ! |
---|
791 | ! ! Migrations |
---|
792 | imigr = jprecj * jpi * jpk * 2 |
---|
793 | ! |
---|
794 | SELECT CASE ( nbondj ) |
---|
795 | CASE ( -1 ) |
---|
796 | CALL mppsend( 4, t4sn(1,1,1,1,1), imigr, nono, ml_req1 ) |
---|
797 | CALL mpprecv( 3, t4ns(1,1,1,1,2), imigr, nono ) |
---|
798 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
799 | CASE ( 0 ) |
---|
800 | CALL mppsend( 3, t4ns(1,1,1,1,1), imigr, noso, ml_req1 ) |
---|
801 | CALL mppsend( 4, t4sn(1,1,1,1,1), imigr, nono, ml_req2 ) |
---|
802 | CALL mpprecv( 3, t4ns(1,1,1,1,2), imigr, nono ) |
---|
803 | CALL mpprecv( 4, t4sn(1,1,1,1,2), imigr, noso ) |
---|
804 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
805 | IF(l_isend) CALL mpi_wait(ml_req2, ml_stat, ml_err) |
---|
806 | CASE ( 1 ) |
---|
807 | CALL mppsend( 3, t4ns(1,1,1,1,1), imigr, noso, ml_req1 ) |
---|
808 | CALL mpprecv( 4, t4sn(1,1,1,1,2), imigr, noso ) |
---|
809 | IF(l_isend) CALL mpi_wait(ml_req1, ml_stat, ml_err) |
---|
810 | END SELECT |
---|
811 | ! |
---|
812 | ! ! Write Dirichlet lateral conditions |
---|
813 | ijhom = nlcj - jprecj |
---|
814 | ! |
---|
815 | SELECT CASE ( nbondj ) |
---|
816 | CASE ( -1 ) |
---|
817 | DO jl = 1, jprecj |
---|
818 | ptab1(:,ijhom+jl,:) = t4ns(:,jl,:,1,2) |
---|
819 | ptab2(:,ijhom+jl,:) = t4ns(:,jl,:,2,2) |
---|
820 | END DO |
---|
821 | CASE ( 0 ) |
---|
822 | DO jl = 1, jprecj |
---|
823 | ptab1(:,jl ,:) = t4sn(:,jl,:,1,2) |
---|
824 | ptab1(:,ijhom+jl,:) = t4ns(:,jl,:,1,2) |
---|
825 | ptab2(:,jl ,:) = t4sn(:,jl,:,2,2) |
---|
826 | ptab2(:,ijhom+jl,:) = t4ns(:,jl,:,2,2) |
---|
827 | END DO |
---|
828 | CASE ( 1 ) |
---|
829 | DO jl = 1, jprecj |
---|
830 | ptab1(:,jl,:) = t4sn(:,jl,:,1,2) |
---|
831 | ptab2(:,jl,:) = t4sn(:,jl,:,2,2) |
---|
832 | END DO |
---|
833 | END SELECT |
---|
834 | |
---|
835 | |
---|
836 | ! 4. north fold treatment |
---|
837 | ! ----------------------- |
---|
838 | IF( npolj /= 0 ) THEN |
---|
839 | ! |
---|
840 | SELECT CASE ( jpni ) |
---|
841 | CASE ( 1 ) |
---|
842 | CALL lbc_nfd ( ptab1, cd_type1, psgn ) ! only for northern procs. |
---|
843 | CALL lbc_nfd ( ptab2, cd_type2, psgn ) |
---|
844 | CASE DEFAULT |
---|
845 | CALL mpp_lbc_north( ptab1, cd_type1, psgn ) ! for all northern procs. |
---|
846 | CALL mpp_lbc_north (ptab2, cd_type2, psgn) |
---|
847 | END SELECT |
---|
848 | ! |
---|
849 | ENDIF |
---|
850 | ! |
---|
851 | END SUBROUTINE mpp_lnk_3d_gather |
---|
852 | |
---|
853 | |
---|
854 | SUBROUTINE mpp_lnk_2d_e( pt2d, cd_type, psgn ) |
---|
855 | !!---------------------------------------------------------------------- |
---|
856 | !! *** routine mpp_lnk_2d_e *** |
---|
857 | !! |
---|
858 | !! ** Purpose : Message passing manadgement for 2d array (with halo) |
---|
859 | !! |
---|
860 | !! ** Method : Use mppsend and mpprecv function for passing mask |
---|
861 | !! between processors following neighboring subdomains. |
---|
862 | !! domain parameters |
---|
863 | !! nlci : first dimension of the local subdomain |
---|
864 | !! nlcj : second dimension of the local subdomain |
---|
865 | !! jpr2di : number of rows for extra outer halo |
---|
866 | !! jpr2dj : number of columns for extra outer halo |
---|
867 | !! nbondi : mark for "east-west local boundary" |
---|
868 | !! nbondj : mark for "north-south local boundary" |
---|
869 | !! noea : number for local neighboring processors |
---|
870 | !! nowe : number for local neighboring processors |
---|
871 | !! noso : number for local neighboring processors |
---|
872 | !! nono : number for local neighboring processors |
---|
873 | !! |
---|
874 | !!---------------------------------------------------------------------- |
---|
875 | REAL(wp), DIMENSION(1-jpr2di:jpi+jpr2di,1-jpr2dj:jpj+jpr2dj), INTENT(inout) :: pt2d ! 2D array with extra halo |
---|
876 | CHARACTER(len=1) , INTENT(in ) :: cd_type ! nature of ptab array grid-points |
---|
877 | ! ! = T , U , V , F , W and I points |
---|
878 | REAL(wp) , INTENT(in ) :: psgn ! =-1 the sign change across the |
---|
879 | !! ! north boundary, = 1. otherwise |
---|
880 | INTEGER :: jl ! dummy loop indices |
---|
881 | INTEGER :: imigr, iihom, ijhom ! temporary integers |
---|
882 | INTEGER :: ipreci, iprecj ! temporary integers |
---|
883 | INTEGER :: ml_req1, ml_req2, ml_err ! for key_mpi_isend |
---|
884 | INTEGER, DIMENSION(MPI_STATUS_SIZE) :: ml_stat ! for key_mpi_isend |
---|
885 | !!---------------------------------------------------------------------- |
---|
886 | |
---|
887 | ipreci = jpreci + jpr2di ! take into account outer extra 2D overlap area |
---|
888 | iprecj = jprecj + jpr2dj |
---|
889 | |
---|
890 | |
---|
891 | ! 1. standard boundary treatment |
---|
892 | ! ------------------------------ |
---|
893 | ! Order matters Here !!!! |
---|
894 | ! |
---|
895 | ! !* North-South boundaries (always colsed) |
---|
896 | IF( .NOT. cd_type == 'F' ) pt2d(:, 1-jpr2dj : jprecj ) = 0.e0 ! south except at F-point |
---|
897 | pt2d(:,nlcj-jprecj+1:jpj+jpr2dj) = 0.e0 ! north |
---|
898 | |
---|
899 | ! ! East-West boundaries |
---|
900 | ! !* Cyclic east-west |
---|
901 | IF( nbondi == 2 .AND. (nperio == 1 .OR. nperio == 4 .OR. nperio == 6) ) THEN |
---|
902 | pt2d(1-jpr2di: 1 ,:) = pt2d(jpim1-jpr2di: jpim1 ,:) ! east |
---|
903 | pt2d( jpi :jpi+jpr2di,:) = pt2d( 2 :2+jpr2di,:) ! west |
---|
904 | ! |
---|
905 | ELSE !* closed |
---|
906 | IF( .NOT. cd_type == 'F' ) pt2d( 1-jpr2di :jpreci ,:) = 0.e0 ! south except at F-point |
---|
907 | pt2d(nlci-jpreci+1:jpi+jpr2di,:) = 0.e0 ! north |
---|
908 | ENDIF |
---|
909 | ! |
---|
910 | |
---|
911 | ! north fold treatment |
---|
912 | ! ----------------------- |
---|
913 | IF( npolj /= 0 ) THEN |
---|
914 | ! |
---|
915 | SELECT CASE ( jpni ) |
---|
916 | CASE ( 1 ) ; CALL lbc_nfd ( pt2d(1:jpi,1:jpj+jpr2dj), cd_type, psgn, pr2dj=jpr2dj ) |
---|
917 | CASE DEFAULT ; CALL mpp_lbc_north_e( pt2d , cd_type, psgn ) |
---|
918 | END SELECT |
---|
919 | ! |
---|
920 | ENDIF |
---|
921 | |
---|
922 | ! 2. East and west directions exchange |
---|
923 | ! ------------------------------------ |
---|
924 | ! we play with the neigbours AND the row number because of the periodicity |
---|
925 | ! |
---|
926 | SELECT CASE ( nbondi ) ! Read Dirichlet lateral conditions |
---|
927 | CASE ( -1, 0, 1 ) ! all exept 2 (i.e. close case) |
---|
928 | iihom = nlci-nreci-jpr2di |
---|
929 | DO jl = 1, ipreci |
---|
930 | tr2ew(:,jl,1) = pt2d(jpreci+jl,:) |
---|
931 | tr2we(:,jl,1) = pt2d(iihom +jl,:) |
---|
932 | END DO |
---|
933 | END SELECT |
---|
934 | ! |
---|
935 | ! ! Migrations |
---|
936 | imigr = ipreci * ( jpj + 2*jpr2dj) |
---|
937 | ! |
---|
938 | SELECT CASE ( nbondi ) |
---|
939 | CASE ( -1 ) |
---|
940 | CALL mppsend( 2, tr2we(1-jpr2dj,1,1), imigr, noea, ml_req1 ) |
---|
941 | CALL mpprecv( 1, tr2ew(1-jpr2dj,1,2), imigr, noea ) |
---|
942 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
943 | CASE ( 0 ) |
---|
944 | CALL mppsend( 1, tr2ew(1-jpr2dj,1,1), imigr, nowe, ml_req1 ) |
---|
945 | CALL mppsend( 2, tr2we(1-jpr2dj,1,1), imigr, noea, ml_req2 ) |
---|
946 | CALL mpprecv( 1, tr2ew(1-jpr2dj,1,2), imigr, noea ) |
---|
947 | CALL mpprecv( 2, tr2we(1-jpr2dj,1,2), imigr, nowe ) |
---|
948 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
949 | IF(l_isend) CALL mpi_wait(ml_req2,ml_stat,ml_err) |
---|
950 | CASE ( 1 ) |
---|
951 | CALL mppsend( 1, tr2ew(1-jpr2dj,1,1), imigr, nowe, ml_req1 ) |
---|
952 | CALL mpprecv( 2, tr2we(1-jpr2dj,1,2), imigr, nowe ) |
---|
953 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
954 | END SELECT |
---|
955 | ! |
---|
956 | ! ! Write Dirichlet lateral conditions |
---|
957 | iihom = nlci - jpreci |
---|
958 | ! |
---|
959 | SELECT CASE ( nbondi ) |
---|
960 | CASE ( -1 ) |
---|
961 | DO jl = 1, ipreci |
---|
962 | pt2d(iihom+jl,:) = tr2ew(:,jl,2) |
---|
963 | END DO |
---|
964 | CASE ( 0 ) |
---|
965 | DO jl = 1, ipreci |
---|
966 | pt2d(jl-jpr2di,:) = tr2we(:,jl,2) |
---|
967 | pt2d( iihom+jl,:) = tr2ew(:,jl,2) |
---|
968 | END DO |
---|
969 | CASE ( 1 ) |
---|
970 | DO jl = 1, ipreci |
---|
971 | pt2d(jl-jpr2di,:) = tr2we(:,jl,2) |
---|
972 | END DO |
---|
973 | END SELECT |
---|
974 | |
---|
975 | |
---|
976 | ! 3. North and south directions |
---|
977 | ! ----------------------------- |
---|
978 | ! always closed : we play only with the neigbours |
---|
979 | ! |
---|
980 | IF( nbondj /= 2 ) THEN ! Read Dirichlet lateral conditions |
---|
981 | ijhom = nlcj-nrecj-jpr2dj |
---|
982 | DO jl = 1, iprecj |
---|
983 | tr2sn(:,jl,1) = pt2d(:,ijhom +jl) |
---|
984 | tr2ns(:,jl,1) = pt2d(:,jprecj+jl) |
---|
985 | END DO |
---|
986 | ENDIF |
---|
987 | ! |
---|
988 | ! ! Migrations |
---|
989 | imigr = iprecj * ( jpi + 2*jpr2di ) |
---|
990 | ! |
---|
991 | SELECT CASE ( nbondj ) |
---|
992 | CASE ( -1 ) |
---|
993 | CALL mppsend( 4, tr2sn(1-jpr2di,1,1), imigr, nono, ml_req1 ) |
---|
994 | CALL mpprecv( 3, tr2ns(1-jpr2di,1,2), imigr, nono ) |
---|
995 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
996 | CASE ( 0 ) |
---|
997 | CALL mppsend( 3, tr2ns(1-jpr2di,1,1), imigr, noso, ml_req1 ) |
---|
998 | CALL mppsend( 4, tr2sn(1-jpr2di,1,1), imigr, nono, ml_req2 ) |
---|
999 | CALL mpprecv( 3, tr2ns(1-jpr2di,1,2), imigr, nono ) |
---|
1000 | CALL mpprecv( 4, tr2sn(1-jpr2di,1,2), imigr, noso ) |
---|
1001 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
1002 | IF(l_isend) CALL mpi_wait(ml_req2,ml_stat,ml_err) |
---|
1003 | CASE ( 1 ) |
---|
1004 | CALL mppsend( 3, tr2ns(1-jpr2di,1,1), imigr, noso, ml_req1 ) |
---|
1005 | CALL mpprecv( 4, tr2sn(1-jpr2di,1,2), imigr, noso ) |
---|
1006 | IF(l_isend) CALL mpi_wait(ml_req1,ml_stat,ml_err) |
---|
1007 | END SELECT |
---|
1008 | ! |
---|
1009 | ! ! Write Dirichlet lateral conditions |
---|
1010 | ijhom = nlcj - jprecj |
---|
1011 | ! |
---|
1012 | SELECT CASE ( nbondj ) |
---|
1013 | CASE ( -1 ) |
---|
1014 | DO jl = 1, iprecj |
---|
1015 | pt2d(:,ijhom+jl) = tr2ns(:,jl,2) |
---|
1016 | END DO |
---|
1017 | CASE ( 0 ) |
---|
1018 | DO jl = 1, iprecj |
---|
1019 | pt2d(:,jl-jpr2dj) = tr2sn(:,jl,2) |
---|
1020 | pt2d(:,ijhom+jl ) = tr2ns(:,jl,2) |
---|
1021 | END DO |
---|
1022 | CASE ( 1 ) |
---|
1023 | DO jl = 1, iprecj |
---|
1024 | pt2d(:,jl-jpr2dj) = tr2sn(:,jl,2) |
---|
1025 | END DO |
---|
1026 | END SELECT |
---|
1027 | |
---|
1028 | END SUBROUTINE mpp_lnk_2d_e |
---|
1029 | |
---|
1030 | |
---|
1031 | SUBROUTINE mppsend( ktyp, pmess, kbytes, kdest, md_req ) |
---|
1032 | !!---------------------------------------------------------------------- |
---|
1033 | !! *** routine mppsend *** |
---|
1034 | !! |
---|
1035 | !! ** Purpose : Send messag passing array |
---|
1036 | !! |
---|
1037 | !!---------------------------------------------------------------------- |
---|
1038 | REAL(wp), INTENT(inout) :: pmess(*) ! array of real |
---|
1039 | INTEGER , INTENT(in ) :: kbytes ! size of the array pmess |
---|
1040 | INTEGER , INTENT(in ) :: kdest ! receive process number |
---|
1041 | INTEGER , INTENT(in ) :: ktyp ! tag of the message |
---|
1042 | INTEGER , INTENT(in ) :: md_req ! argument for isend |
---|
1043 | !! |
---|
1044 | INTEGER :: iflag |
---|
1045 | !!---------------------------------------------------------------------- |
---|
1046 | ! |
---|
1047 | SELECT CASE ( cn_mpi_send ) |
---|
1048 | CASE ( 'S' ) ! Standard mpi send (blocking) |
---|
1049 | CALL mpi_send ( pmess, kbytes, mpi_double_precision, kdest , ktyp, mpi_comm_opa , iflag ) |
---|
1050 | CASE ( 'B' ) ! Buffer mpi send (blocking) |
---|
1051 | CALL mpi_bsend( pmess, kbytes, mpi_double_precision, kdest , ktyp, mpi_comm_opa , iflag ) |
---|
1052 | CASE ( 'I' ) ! Immediate mpi send (non-blocking send) |
---|
1053 | ! be carefull, one more argument here : the mpi request identifier.. |
---|
1054 | CALL mpi_isend( pmess, kbytes, mpi_double_precision, kdest , ktyp, mpi_comm_opa, md_req, iflag ) |
---|
1055 | END SELECT |
---|
1056 | ! |
---|
1057 | END SUBROUTINE mppsend |
---|
1058 | |
---|
1059 | |
---|
1060 | SUBROUTINE mpprecv( ktyp, pmess, kbytes, ksource ) |
---|
1061 | !!---------------------------------------------------------------------- |
---|
1062 | !! *** routine mpprecv *** |
---|
1063 | !! |
---|
1064 | !! ** Purpose : Receive messag passing array |
---|
1065 | !! |
---|
1066 | !!---------------------------------------------------------------------- |
---|
1067 | REAL(wp), INTENT(inout) :: pmess(*) ! array of real |
---|
1068 | INTEGER , INTENT(in ) :: kbytes ! suze of the array pmess |
---|
1069 | INTEGER , INTENT(in ) :: ktyp ! Tag of the recevied message |
---|
1070 | INTEGER, OPTIONAL, INTENT(in) :: ksource ! source process number |
---|
1071 | !! |
---|
1072 | INTEGER :: istatus(mpi_status_size) |
---|
1073 | INTEGER :: iflag |
---|
1074 | INTEGER :: use_source |
---|
1075 | !!---------------------------------------------------------------------- |
---|
1076 | ! |
---|
1077 | |
---|
1078 | ! If a specific process number has been passed to the receive call, |
---|
1079 | ! use that one. Default is to use mpi_any_source |
---|
1080 | use_source=mpi_any_source |
---|
1081 | if(present(ksource)) then |
---|
1082 | use_source=ksource |
---|
1083 | end if |
---|
1084 | |
---|
1085 | CALL mpi_recv( pmess, kbytes, mpi_double_precision, use_source, ktyp, mpi_comm_opa, istatus, iflag ) |
---|
1086 | ! |
---|
1087 | END SUBROUTINE mpprecv |
---|
1088 | |
---|
1089 | |
---|
1090 | SUBROUTINE mppgather( ptab, kp, pio ) |
---|
1091 | !!---------------------------------------------------------------------- |
---|
1092 | !! *** routine mppgather *** |
---|
1093 | !! |
---|
1094 | !! ** Purpose : Transfert between a local subdomain array and a work |
---|
1095 | !! array which is distributed following the vertical level. |
---|
1096 | !! |
---|
1097 | !!---------------------------------------------------------------------- |
---|
1098 | REAL(wp), DIMENSION(jpi,jpj), INTENT(in ) :: ptab ! subdomain input array |
---|
1099 | INTEGER , INTENT(in ) :: kp ! record length |
---|
1100 | REAL(wp), DIMENSION(jpi,jpj,jpnij), INTENT( out) :: pio ! subdomain input array |
---|
1101 | !! |
---|
1102 | INTEGER :: itaille, ierror ! temporary integer |
---|
1103 | !!--------------------------------------------------------------------- |
---|
1104 | ! |
---|
1105 | itaille = jpi * jpj |
---|
1106 | CALL mpi_gather( ptab, itaille, mpi_double_precision, pio, itaille , & |
---|
1107 | & mpi_double_precision, kp , mpi_comm_opa, ierror ) |
---|
1108 | ! |
---|
1109 | END SUBROUTINE mppgather |
---|
1110 | |
---|
1111 | |
---|
1112 | SUBROUTINE mppscatter( pio, kp, ptab ) |
---|
1113 | !!---------------------------------------------------------------------- |
---|
1114 | !! *** routine mppscatter *** |
---|
1115 | !! |
---|
1116 | !! ** Purpose : Transfert between awork array which is distributed |
---|
1117 | !! following the vertical level and the local subdomain array. |
---|
1118 | !! |
---|
1119 | !!---------------------------------------------------------------------- |
---|
1120 | REAL(wp), DIMENSION(jpi,jpj,jpnij) :: pio ! output array |
---|
1121 | INTEGER :: kp ! Tag (not used with MPI |
---|
1122 | REAL(wp), DIMENSION(jpi,jpj) :: ptab ! subdomain array input |
---|
1123 | !! |
---|
1124 | INTEGER :: itaille, ierror ! temporary integer |
---|
1125 | !!--------------------------------------------------------------------- |
---|
1126 | ! |
---|
1127 | itaille=jpi*jpj |
---|
1128 | ! |
---|
1129 | CALL mpi_scatter( pio, itaille, mpi_double_precision, ptab, itaille , & |
---|
1130 | & mpi_double_precision, kp , mpi_comm_opa, ierror ) |
---|
1131 | ! |
---|
1132 | END SUBROUTINE mppscatter |
---|
1133 | |
---|
1134 | |
---|
1135 | SUBROUTINE mppmax_a_int( ktab, kdim, kcom ) |
---|
1136 | !!---------------------------------------------------------------------- |
---|
1137 | !! *** routine mppmax_a_int *** |
---|
1138 | !! |
---|
1139 | !! ** Purpose : Find maximum value in an integer layout array |
---|
1140 | !! |
---|
1141 | !!---------------------------------------------------------------------- |
---|
1142 | INTEGER , INTENT(in ) :: kdim ! size of array |
---|
1143 | INTEGER , INTENT(inout), DIMENSION(kdim) :: ktab ! input array |
---|
1144 | INTEGER , INTENT(in ), OPTIONAL :: kcom ! |
---|
1145 | !! |
---|
1146 | INTEGER :: ierror, localcomm ! temporary integer |
---|
1147 | INTEGER, DIMENSION(kdim) :: iwork |
---|
1148 | !!---------------------------------------------------------------------- |
---|
1149 | ! |
---|
1150 | localcomm = mpi_comm_opa |
---|
1151 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1152 | ! |
---|
1153 | CALL mpi_allreduce( ktab, iwork, kdim, mpi_integer, mpi_max, localcomm, ierror ) |
---|
1154 | ! |
---|
1155 | ktab(:) = iwork(:) |
---|
1156 | ! |
---|
1157 | END SUBROUTINE mppmax_a_int |
---|
1158 | |
---|
1159 | |
---|
1160 | SUBROUTINE mppmax_int( ktab, kcom ) |
---|
1161 | !!---------------------------------------------------------------------- |
---|
1162 | !! *** routine mppmax_int *** |
---|
1163 | !! |
---|
1164 | !! ** Purpose : Find maximum value in an integer layout array |
---|
1165 | !! |
---|
1166 | !!---------------------------------------------------------------------- |
---|
1167 | INTEGER, INTENT(inout) :: ktab ! ??? |
---|
1168 | INTEGER, INTENT(in ), OPTIONAL :: kcom ! ??? |
---|
1169 | !! |
---|
1170 | INTEGER :: ierror, iwork, localcomm ! temporary integer |
---|
1171 | !!---------------------------------------------------------------------- |
---|
1172 | ! |
---|
1173 | localcomm = mpi_comm_opa |
---|
1174 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1175 | ! |
---|
1176 | CALL mpi_allreduce( ktab, iwork, 1, mpi_integer, mpi_max, localcomm, ierror) |
---|
1177 | ! |
---|
1178 | ktab = iwork |
---|
1179 | ! |
---|
1180 | END SUBROUTINE mppmax_int |
---|
1181 | |
---|
1182 | |
---|
1183 | SUBROUTINE mppmin_a_int( ktab, kdim, kcom ) |
---|
1184 | !!---------------------------------------------------------------------- |
---|
1185 | !! *** routine mppmin_a_int *** |
---|
1186 | !! |
---|
1187 | !! ** Purpose : Find minimum value in an integer layout array |
---|
1188 | !! |
---|
1189 | !!---------------------------------------------------------------------- |
---|
1190 | INTEGER , INTENT( in ) :: kdim ! size of array |
---|
1191 | INTEGER , INTENT(inout), DIMENSION(kdim) :: ktab ! input array |
---|
1192 | INTEGER , INTENT( in ), OPTIONAL :: kcom ! input array |
---|
1193 | !! |
---|
1194 | INTEGER :: ierror, localcomm ! temporary integer |
---|
1195 | INTEGER, DIMENSION(kdim) :: iwork |
---|
1196 | !!---------------------------------------------------------------------- |
---|
1197 | ! |
---|
1198 | localcomm = mpi_comm_opa |
---|
1199 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1200 | ! |
---|
1201 | CALL mpi_allreduce( ktab, iwork, kdim, mpi_integer, mpi_min, localcomm, ierror ) |
---|
1202 | ! |
---|
1203 | ktab(:) = iwork(:) |
---|
1204 | ! |
---|
1205 | END SUBROUTINE mppmin_a_int |
---|
1206 | |
---|
1207 | |
---|
1208 | SUBROUTINE mppmin_int( ktab, kcom ) |
---|
1209 | !!---------------------------------------------------------------------- |
---|
1210 | !! *** routine mppmin_int *** |
---|
1211 | !! |
---|
1212 | !! ** Purpose : Find minimum value in an integer layout array |
---|
1213 | !! |
---|
1214 | !!---------------------------------------------------------------------- |
---|
1215 | INTEGER, INTENT(inout) :: ktab ! ??? |
---|
1216 | INTEGER , INTENT( in ), OPTIONAL :: kcom ! input array |
---|
1217 | !! |
---|
1218 | INTEGER :: ierror, iwork, localcomm |
---|
1219 | !!---------------------------------------------------------------------- |
---|
1220 | ! |
---|
1221 | localcomm = mpi_comm_opa |
---|
1222 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1223 | ! |
---|
1224 | CALL mpi_allreduce( ktab, iwork, 1, mpi_integer, mpi_min, localcomm, ierror ) |
---|
1225 | ! |
---|
1226 | ktab = iwork |
---|
1227 | ! |
---|
1228 | END SUBROUTINE mppmin_int |
---|
1229 | |
---|
1230 | |
---|
1231 | SUBROUTINE mppsum_a_int( ktab, kdim ) |
---|
1232 | !!---------------------------------------------------------------------- |
---|
1233 | !! *** routine mppsum_a_int *** |
---|
1234 | !! |
---|
1235 | !! ** Purpose : Global integer sum, 1D array case |
---|
1236 | !! |
---|
1237 | !!---------------------------------------------------------------------- |
---|
1238 | INTEGER, INTENT(in ) :: kdim ! ??? |
---|
1239 | INTEGER, INTENT(inout), DIMENSION (kdim) :: ktab ! ??? |
---|
1240 | !! |
---|
1241 | INTEGER :: ierror |
---|
1242 | INTEGER, DIMENSION (kdim) :: iwork |
---|
1243 | !!---------------------------------------------------------------------- |
---|
1244 | ! |
---|
1245 | CALL mpi_allreduce( ktab, iwork, kdim, mpi_integer, mpi_sum, mpi_comm_opa, ierror ) |
---|
1246 | ! |
---|
1247 | ktab(:) = iwork(:) |
---|
1248 | ! |
---|
1249 | END SUBROUTINE mppsum_a_int |
---|
1250 | |
---|
1251 | |
---|
1252 | SUBROUTINE mppsum_int( ktab ) |
---|
1253 | !!---------------------------------------------------------------------- |
---|
1254 | !! *** routine mppsum_int *** |
---|
1255 | !! |
---|
1256 | !! ** Purpose : Global integer sum |
---|
1257 | !! |
---|
1258 | !!---------------------------------------------------------------------- |
---|
1259 | INTEGER, INTENT(inout) :: ktab |
---|
1260 | !! |
---|
1261 | INTEGER :: ierror, iwork |
---|
1262 | !!---------------------------------------------------------------------- |
---|
1263 | ! |
---|
1264 | CALL mpi_allreduce( ktab, iwork, 1, mpi_integer, mpi_sum, mpi_comm_opa, ierror ) |
---|
1265 | ! |
---|
1266 | ktab = iwork |
---|
1267 | ! |
---|
1268 | END SUBROUTINE mppsum_int |
---|
1269 | |
---|
1270 | |
---|
1271 | SUBROUTINE mppmax_a_real( ptab, kdim, kcom ) |
---|
1272 | !!---------------------------------------------------------------------- |
---|
1273 | !! *** routine mppmax_a_real *** |
---|
1274 | !! |
---|
1275 | !! ** Purpose : Maximum |
---|
1276 | !! |
---|
1277 | !!---------------------------------------------------------------------- |
---|
1278 | INTEGER , INTENT(in ) :: kdim |
---|
1279 | REAL(wp), INTENT(inout), DIMENSION(kdim) :: ptab |
---|
1280 | INTEGER , INTENT(in ), OPTIONAL :: kcom |
---|
1281 | !! |
---|
1282 | INTEGER :: ierror, localcomm |
---|
1283 | REAL(wp), DIMENSION(kdim) :: zwork |
---|
1284 | !!---------------------------------------------------------------------- |
---|
1285 | ! |
---|
1286 | localcomm = mpi_comm_opa |
---|
1287 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1288 | ! |
---|
1289 | CALL mpi_allreduce( ptab, zwork, kdim, mpi_double_precision, mpi_max, localcomm, ierror ) |
---|
1290 | ptab(:) = zwork(:) |
---|
1291 | ! |
---|
1292 | END SUBROUTINE mppmax_a_real |
---|
1293 | |
---|
1294 | |
---|
1295 | SUBROUTINE mppmax_real( ptab, kcom ) |
---|
1296 | !!---------------------------------------------------------------------- |
---|
1297 | !! *** routine mppmax_real *** |
---|
1298 | !! |
---|
1299 | !! ** Purpose : Maximum |
---|
1300 | !! |
---|
1301 | !!---------------------------------------------------------------------- |
---|
1302 | REAL(wp), INTENT(inout) :: ptab ! ??? |
---|
1303 | INTEGER , INTENT(in ), OPTIONAL :: kcom ! ??? |
---|
1304 | !! |
---|
1305 | INTEGER :: ierror, localcomm |
---|
1306 | REAL(wp) :: zwork |
---|
1307 | !!---------------------------------------------------------------------- |
---|
1308 | ! |
---|
1309 | localcomm = mpi_comm_opa |
---|
1310 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1311 | ! |
---|
1312 | CALL mpi_allreduce( ptab, zwork, 1, mpi_double_precision, mpi_max, localcomm, ierror ) |
---|
1313 | ptab = zwork |
---|
1314 | ! |
---|
1315 | END SUBROUTINE mppmax_real |
---|
1316 | |
---|
1317 | |
---|
1318 | SUBROUTINE mppmin_a_real( ptab, kdim, kcom ) |
---|
1319 | !!---------------------------------------------------------------------- |
---|
1320 | !! *** routine mppmin_a_real *** |
---|
1321 | !! |
---|
1322 | !! ** Purpose : Minimum of REAL, array case |
---|
1323 | !! |
---|
1324 | !!----------------------------------------------------------------------- |
---|
1325 | INTEGER , INTENT(in ) :: kdim |
---|
1326 | REAL(wp), INTENT(inout), DIMENSION(kdim) :: ptab |
---|
1327 | INTEGER , INTENT(in ), OPTIONAL :: kcom |
---|
1328 | !! |
---|
1329 | INTEGER :: ierror, localcomm |
---|
1330 | REAL(wp), DIMENSION(kdim) :: zwork |
---|
1331 | !!----------------------------------------------------------------------- |
---|
1332 | ! |
---|
1333 | localcomm = mpi_comm_opa |
---|
1334 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1335 | ! |
---|
1336 | CALL mpi_allreduce( ptab, zwork, kdim, mpi_double_precision, mpi_min, localcomm, ierror ) |
---|
1337 | ptab(:) = zwork(:) |
---|
1338 | ! |
---|
1339 | END SUBROUTINE mppmin_a_real |
---|
1340 | |
---|
1341 | |
---|
1342 | SUBROUTINE mppmin_real( ptab, kcom ) |
---|
1343 | !!---------------------------------------------------------------------- |
---|
1344 | !! *** routine mppmin_real *** |
---|
1345 | !! |
---|
1346 | !! ** Purpose : minimum of REAL, scalar case |
---|
1347 | !! |
---|
1348 | !!----------------------------------------------------------------------- |
---|
1349 | REAL(wp), INTENT(inout) :: ptab ! |
---|
1350 | INTEGER , INTENT(in ), OPTIONAL :: kcom |
---|
1351 | !! |
---|
1352 | INTEGER :: ierror |
---|
1353 | REAL(wp) :: zwork |
---|
1354 | INTEGER :: localcomm |
---|
1355 | !!----------------------------------------------------------------------- |
---|
1356 | ! |
---|
1357 | localcomm = mpi_comm_opa |
---|
1358 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1359 | ! |
---|
1360 | CALL mpi_allreduce( ptab, zwork, 1, mpi_double_precision, mpi_min, localcomm, ierror ) |
---|
1361 | ptab = zwork |
---|
1362 | ! |
---|
1363 | END SUBROUTINE mppmin_real |
---|
1364 | |
---|
1365 | |
---|
1366 | SUBROUTINE mppsum_a_real( ptab, kdim, kcom ) |
---|
1367 | !!---------------------------------------------------------------------- |
---|
1368 | !! *** routine mppsum_a_real *** |
---|
1369 | !! |
---|
1370 | !! ** Purpose : global sum, REAL ARRAY argument case |
---|
1371 | !! |
---|
1372 | !!----------------------------------------------------------------------- |
---|
1373 | INTEGER , INTENT( in ) :: kdim ! size of ptab |
---|
1374 | REAL(wp), DIMENSION(kdim), INTENT( inout ) :: ptab ! input array |
---|
1375 | INTEGER , INTENT( in ), OPTIONAL :: kcom |
---|
1376 | !! |
---|
1377 | INTEGER :: ierror ! temporary integer |
---|
1378 | INTEGER :: localcomm |
---|
1379 | REAL(wp), DIMENSION(kdim) :: zwork ! temporary workspace |
---|
1380 | !!----------------------------------------------------------------------- |
---|
1381 | ! |
---|
1382 | localcomm = mpi_comm_opa |
---|
1383 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1384 | ! |
---|
1385 | CALL mpi_allreduce( ptab, zwork, kdim, mpi_double_precision, mpi_sum, localcomm, ierror ) |
---|
1386 | ptab(:) = zwork(:) |
---|
1387 | ! |
---|
1388 | END SUBROUTINE mppsum_a_real |
---|
1389 | |
---|
1390 | |
---|
1391 | SUBROUTINE mppsum_real( ptab, kcom ) |
---|
1392 | !!---------------------------------------------------------------------- |
---|
1393 | !! *** routine mppsum_real *** |
---|
1394 | !! |
---|
1395 | !! ** Purpose : global sum, SCALAR argument case |
---|
1396 | !! |
---|
1397 | !!----------------------------------------------------------------------- |
---|
1398 | REAL(wp), INTENT(inout) :: ptab ! input scalar |
---|
1399 | INTEGER , INTENT(in ), OPTIONAL :: kcom |
---|
1400 | !! |
---|
1401 | INTEGER :: ierror, localcomm |
---|
1402 | REAL(wp) :: zwork |
---|
1403 | !!----------------------------------------------------------------------- |
---|
1404 | ! |
---|
1405 | localcomm = mpi_comm_opa |
---|
1406 | IF( PRESENT(kcom) ) localcomm = kcom |
---|
1407 | ! |
---|
1408 | CALL mpi_allreduce( ptab, zwork, 1, mpi_double_precision, mpi_sum, localcomm, ierror ) |
---|
1409 | ptab = zwork |
---|
1410 | ! |
---|
1411 | END SUBROUTINE mppsum_real |
---|
1412 | |
---|
1413 | |
---|
1414 | SUBROUTINE mpp_minloc2d( ptab, pmask, pmin, ki,kj ) |
---|
1415 | !!------------------------------------------------------------------------ |
---|
1416 | !! *** routine mpp_minloc *** |
---|
1417 | !! |
---|
1418 | !! ** Purpose : Compute the global minimum of an array ptab |
---|
1419 | !! and also give its global position |
---|
1420 | !! |
---|
1421 | !! ** Method : Use MPI_ALLREDUCE with MPI_MINLOC |
---|
1422 | !! |
---|
1423 | !!-------------------------------------------------------------------------- |
---|
1424 | REAL(wp), DIMENSION (jpi,jpj), INTENT(in ) :: ptab ! Local 2D array |
---|
1425 | REAL(wp), DIMENSION (jpi,jpj), INTENT(in ) :: pmask ! Local mask |
---|
1426 | REAL(wp) , INTENT( out) :: pmin ! Global minimum of ptab |
---|
1427 | INTEGER , INTENT( out) :: ki, kj ! index of minimum in global frame |
---|
1428 | !! |
---|
1429 | INTEGER , DIMENSION(2) :: ilocs |
---|
1430 | INTEGER :: ierror |
---|
1431 | REAL(wp) :: zmin ! local minimum |
---|
1432 | REAL(wp), DIMENSION(2,1) :: zain, zaout |
---|
1433 | !!----------------------------------------------------------------------- |
---|
1434 | ! |
---|
1435 | zmin = MINVAL( ptab(:,:) , mask= pmask == 1.e0 ) |
---|
1436 | ilocs = MINLOC( ptab(:,:) , mask= pmask == 1.e0 ) |
---|
1437 | ! |
---|
1438 | ki = ilocs(1) + nimpp - 1 |
---|
1439 | kj = ilocs(2) + njmpp - 1 |
---|
1440 | ! |
---|
1441 | zain(1,:)=zmin |
---|
1442 | zain(2,:)=ki+10000.*kj |
---|
1443 | ! |
---|
1444 | CALL MPI_ALLREDUCE( zain,zaout, 1, MPI_2DOUBLE_PRECISION,MPI_MINLOC,MPI_COMM_OPA,ierror) |
---|
1445 | ! |
---|
1446 | pmin = zaout(1,1) |
---|
1447 | kj = INT(zaout(2,1)/10000.) |
---|
1448 | ki = INT(zaout(2,1) - 10000.*kj ) |
---|
1449 | ! |
---|
1450 | END SUBROUTINE mpp_minloc2d |
---|
1451 | |
---|
1452 | |
---|
1453 | SUBROUTINE mpp_minloc3d( ptab, pmask, pmin, ki, kj ,kk) |
---|
1454 | !!------------------------------------------------------------------------ |
---|
1455 | !! *** routine mpp_minloc *** |
---|
1456 | !! |
---|
1457 | !! ** Purpose : Compute the global minimum of an array ptab |
---|
1458 | !! and also give its global position |
---|
1459 | !! |
---|
1460 | !! ** Method : Use MPI_ALLREDUCE with MPI_MINLOC |
---|
1461 | !! |
---|
1462 | !!-------------------------------------------------------------------------- |
---|
1463 | REAL(wp), DIMENSION (jpi,jpj,jpk), INTENT(in ) :: ptab ! Local 2D array |
---|
1464 | REAL(wp), DIMENSION (jpi,jpj,jpk), INTENT(in ) :: pmask ! Local mask |
---|
1465 | REAL(wp) , INTENT( out) :: pmin ! Global minimum of ptab |
---|
1466 | INTEGER , INTENT( out) :: ki, kj, kk ! index of minimum in global frame |
---|
1467 | !! |
---|
1468 | INTEGER :: ierror |
---|
1469 | REAL(wp) :: zmin ! local minimum |
---|
1470 | INTEGER , DIMENSION(3) :: ilocs |
---|
1471 | REAL(wp), DIMENSION(2,1) :: zain, zaout |
---|
1472 | !!----------------------------------------------------------------------- |
---|
1473 | ! |
---|
1474 | zmin = MINVAL( ptab(:,:,:) , mask= pmask == 1.e0 ) |
---|
1475 | ilocs = MINLOC( ptab(:,:,:) , mask= pmask == 1.e0 ) |
---|
1476 | ! |
---|
1477 | ki = ilocs(1) + nimpp - 1 |
---|
1478 | kj = ilocs(2) + njmpp - 1 |
---|
1479 | kk = ilocs(3) |
---|
1480 | ! |
---|
1481 | zain(1,:)=zmin |
---|
1482 | zain(2,:)=ki+10000.*kj+100000000.*kk |
---|
1483 | ! |
---|
1484 | CALL MPI_ALLREDUCE( zain,zaout, 1, MPI_2DOUBLE_PRECISION,MPI_MINLOC,MPI_COMM_OPA,ierror) |
---|
1485 | ! |
---|
1486 | pmin = zaout(1,1) |
---|
1487 | kk = INT( zaout(2,1) / 100000000. ) |
---|
1488 | kj = INT( zaout(2,1) - kk * 100000000. ) / 10000 |
---|
1489 | ki = INT( zaout(2,1) - kk * 100000000. -kj * 10000. ) |
---|
1490 | ! |
---|
1491 | END SUBROUTINE mpp_minloc3d |
---|
1492 | |
---|
1493 | |
---|
1494 | SUBROUTINE mpp_maxloc2d( ptab, pmask, pmax, ki, kj ) |
---|
1495 | !!------------------------------------------------------------------------ |
---|
1496 | !! *** routine mpp_maxloc *** |
---|
1497 | !! |
---|
1498 | !! ** Purpose : Compute the global maximum of an array ptab |
---|
1499 | !! and also give its global position |
---|
1500 | !! |
---|
1501 | !! ** Method : Use MPI_ALLREDUCE with MPI_MINLOC |
---|
1502 | !! |
---|
1503 | !!-------------------------------------------------------------------------- |
---|
1504 | REAL(wp), DIMENSION (jpi,jpj), INTENT(in ) :: ptab ! Local 2D array |
---|
1505 | REAL(wp), DIMENSION (jpi,jpj), INTENT(in ) :: pmask ! Local mask |
---|
1506 | REAL(wp) , INTENT( out) :: pmax ! Global maximum of ptab |
---|
1507 | INTEGER , INTENT( out) :: ki, kj ! index of maximum in global frame |
---|
1508 | !! |
---|
1509 | INTEGER :: ierror |
---|
1510 | INTEGER, DIMENSION (2) :: ilocs |
---|
1511 | REAL(wp) :: zmax ! local maximum |
---|
1512 | REAL(wp), DIMENSION(2,1) :: zain, zaout |
---|
1513 | !!----------------------------------------------------------------------- |
---|
1514 | ! |
---|
1515 | zmax = MAXVAL( ptab(:,:) , mask= pmask == 1.e0 ) |
---|
1516 | ilocs = MAXLOC( ptab(:,:) , mask= pmask == 1.e0 ) |
---|
1517 | ! |
---|
1518 | ki = ilocs(1) + nimpp - 1 |
---|
1519 | kj = ilocs(2) + njmpp - 1 |
---|
1520 | ! |
---|
1521 | zain(1,:) = zmax |
---|
1522 | zain(2,:) = ki + 10000. * kj |
---|
1523 | ! |
---|
1524 | CALL MPI_ALLREDUCE( zain,zaout, 1, MPI_2DOUBLE_PRECISION,MPI_MAXLOC,MPI_COMM_OPA,ierror) |
---|
1525 | ! |
---|
1526 | pmax = zaout(1,1) |
---|
1527 | kj = INT( zaout(2,1) / 10000. ) |
---|
1528 | ki = INT( zaout(2,1) - 10000.* kj ) |
---|
1529 | ! |
---|
1530 | END SUBROUTINE mpp_maxloc2d |
---|
1531 | |
---|
1532 | |
---|
1533 | SUBROUTINE mpp_maxloc3d( ptab, pmask, pmax, ki, kj, kk ) |
---|
1534 | !!------------------------------------------------------------------------ |
---|
1535 | !! *** routine mpp_maxloc *** |
---|
1536 | !! |
---|
1537 | !! ** Purpose : Compute the global maximum of an array ptab |
---|
1538 | !! and also give its global position |
---|
1539 | !! |
---|
1540 | !! ** Method : Use MPI_ALLREDUCE with MPI_MINLOC |
---|
1541 | !! |
---|
1542 | !!-------------------------------------------------------------------------- |
---|
1543 | REAL(wp), DIMENSION (jpi,jpj,jpk), INTENT(in ) :: ptab ! Local 2D array |
---|
1544 | REAL(wp), DIMENSION (jpi,jpj,jpk), INTENT(in ) :: pmask ! Local mask |
---|
1545 | REAL(wp) , INTENT( out) :: pmax ! Global maximum of ptab |
---|
1546 | INTEGER , INTENT( out) :: ki, kj, kk ! index of maximum in global frame |
---|
1547 | !! |
---|
1548 | REAL(wp) :: zmax ! local maximum |
---|
1549 | REAL(wp), DIMENSION(2,1) :: zain, zaout |
---|
1550 | INTEGER , DIMENSION(3) :: ilocs |
---|
1551 | INTEGER :: ierror |
---|
1552 | !!----------------------------------------------------------------------- |
---|
1553 | ! |
---|
1554 | zmax = MAXVAL( ptab(:,:,:) , mask= pmask == 1.e0 ) |
---|
1555 | ilocs = MAXLOC( ptab(:,:,:) , mask= pmask == 1.e0 ) |
---|
1556 | ! |
---|
1557 | ki = ilocs(1) + nimpp - 1 |
---|
1558 | kj = ilocs(2) + njmpp - 1 |
---|
1559 | kk = ilocs(3) |
---|
1560 | ! |
---|
1561 | zain(1,:)=zmax |
---|
1562 | zain(2,:)=ki+10000.*kj+100000000.*kk |
---|
1563 | ! |
---|
1564 | CALL MPI_ALLREDUCE( zain,zaout, 1, MPI_2DOUBLE_PRECISION,MPI_MAXLOC,MPI_COMM_OPA,ierror) |
---|
1565 | ! |
---|
1566 | pmax = zaout(1,1) |
---|
1567 | kk = INT( zaout(2,1) / 100000000. ) |
---|
1568 | kj = INT( zaout(2,1) - kk * 100000000. ) / 10000 |
---|
1569 | ki = INT( zaout(2,1) - kk * 100000000. -kj * 10000. ) |
---|
1570 | ! |
---|
1571 | END SUBROUTINE mpp_maxloc3d |
---|
1572 | |
---|
1573 | |
---|
1574 | SUBROUTINE mppsync() |
---|
1575 | !!---------------------------------------------------------------------- |
---|
1576 | !! *** routine mppsync *** |
---|
1577 | !! |
---|
1578 | !! ** Purpose : Massively parallel processors, synchroneous |
---|
1579 | !! |
---|
1580 | !!----------------------------------------------------------------------- |
---|
1581 | INTEGER :: ierror |
---|
1582 | !!----------------------------------------------------------------------- |
---|
1583 | ! |
---|
1584 | CALL mpi_barrier( mpi_comm_opa, ierror ) |
---|
1585 | ! |
---|
1586 | END SUBROUTINE mppsync |
---|
1587 | |
---|
1588 | |
---|
1589 | SUBROUTINE mppstop |
---|
1590 | !!---------------------------------------------------------------------- |
---|
1591 | !! *** routine mppstop *** |
---|
1592 | !! |
---|
1593 | !! ** purpose : Stop massilively parallel processors method |
---|
1594 | !! |
---|
1595 | !!---------------------------------------------------------------------- |
---|
1596 | INTEGER :: info |
---|
1597 | !!---------------------------------------------------------------------- |
---|
1598 | ! |
---|
1599 | CALL mppsync |
---|
1600 | CALL mpi_finalize( info ) |
---|
1601 | ! |
---|
1602 | END SUBROUTINE mppstop |
---|
1603 | |
---|
1604 | |
---|
1605 | SUBROUTINE mppobc( ptab, kd1, kd2, kl, kk, ktype, kij ) |
---|
1606 | !!---------------------------------------------------------------------- |
---|
1607 | !! *** routine mppobc *** |
---|
1608 | !! |
---|
1609 | !! ** Purpose : Message passing manadgement for open boundary |
---|
1610 | !! conditions array |
---|
1611 | !! |
---|
1612 | !! ** Method : Use mppsend and mpprecv function for passing mask |
---|
1613 | !! between processors following neighboring subdomains. |
---|
1614 | !! domain parameters |
---|
1615 | !! nlci : first dimension of the local subdomain |
---|
1616 | !! nlcj : second dimension of the local subdomain |
---|
1617 | !! nbondi : mark for "east-west local boundary" |
---|
1618 | !! nbondj : mark for "north-south local boundary" |
---|
1619 | !! noea : number for local neighboring processors |
---|
1620 | !! nowe : number for local neighboring processors |
---|
1621 | !! noso : number for local neighboring processors |
---|
1622 | !! nono : number for local neighboring processors |
---|
1623 | !! |
---|
1624 | !!---------------------------------------------------------------------- |
---|
1625 | INTEGER , INTENT(in ) :: kd1, kd2 ! starting and ending indices |
---|
1626 | INTEGER , INTENT(in ) :: kl ! index of open boundary |
---|
1627 | INTEGER , INTENT(in ) :: kk ! vertical dimension |
---|
1628 | INTEGER , INTENT(in ) :: ktype ! define north/south or east/west cdt |
---|
1629 | ! ! = 1 north/south ; = 2 east/west |
---|
1630 | INTEGER , INTENT(in ) :: kij ! horizontal dimension |
---|
1631 | REAL(wp), INTENT(inout), DIMENSION(kij,kk) :: ptab ! variable array |
---|
1632 | !! |
---|
1633 | INTEGER :: ji, jj, jk, jl ! dummy loop indices |
---|
1634 | INTEGER :: iipt0, iipt1, ilpt1 ! temporary integers |
---|
1635 | INTEGER :: ijpt0, ijpt1 ! - - |
---|
1636 | INTEGER :: imigr, iihom, ijhom ! - - |
---|
1637 | INTEGER :: ml_req1, ml_req2, ml_err ! for key_mpi_isend |
---|
1638 | INTEGER :: ml_stat(MPI_STATUS_SIZE) ! for key_mpi_isend |
---|
1639 | REAL(wp), DIMENSION(jpi,jpj) :: ztab ! temporary workspace |
---|
1640 | !!---------------------------------------------------------------------- |
---|
1641 | |
---|
1642 | ! boundary condition initialization |
---|
1643 | ! --------------------------------- |
---|
1644 | ztab(:,:) = 0.e0 |
---|
1645 | ! |
---|
1646 | IF( ktype==1 ) THEN ! north/south boundaries |
---|
1647 | iipt0 = MAX( 1, MIN(kd1 - nimpp+1, nlci ) ) |
---|
1648 | iipt1 = MAX( 0, MIN(kd2 - nimpp+1, nlci - 1 ) ) |
---|
1649 | ilpt1 = MAX( 1, MIN(kd2 - nimpp+1, nlci ) ) |
---|
1650 | ijpt0 = MAX( 1, MIN(kl - njmpp+1, nlcj ) ) |
---|
1651 | ijpt1 = MAX( 0, MIN(kl - njmpp+1, nlcj - 1 ) ) |
---|
1652 | ELSEIF( ktype==2 ) THEN ! east/west boundaries |
---|
1653 | iipt0 = MAX( 1, MIN(kl - nimpp+1, nlci ) ) |
---|
1654 | iipt1 = MAX( 0, MIN(kl - nimpp+1, nlci - 1 ) ) |
---|
1655 | ijpt0 = MAX( 1, MIN(kd1 - njmpp+1, nlcj ) ) |
---|
1656 | ijpt1 = MAX( 0, MIN(kd2 - njmpp+1, nlcj - 1 ) ) |
---|
1657 | ilpt1 = MAX( 1, MIN(kd2 - njmpp+1, nlcj ) ) |
---|
1658 | ELSE |
---|
1659 | CALL ctl_stop( 'mppobc: bad ktype' ) |
---|
1660 | ENDIF |
---|
1661 | |
---|
1662 | ! Communication level by level |
---|
1663 | ! ---------------------------- |
---|
1664 | !!gm Remark : this is very time consumming!!! |
---|
1665 | ! ! ------------------------ ! |
---|
1666 | DO jk = 1, kk ! Loop over the levels ! |
---|
1667 | ! ! ------------------------ ! |
---|
1668 | ! |
---|
1669 | IF( ktype == 1 ) THEN ! north/south boundaries |
---|
1670 | DO jj = ijpt0, ijpt1 |
---|
1671 | DO ji = iipt0, iipt1 |
---|
1672 | ztab(ji,jj) = ptab(ji,jk) |
---|
1673 | END DO |
---|
1674 | END DO |
---|
1675 | ELSEIF( ktype == 2 ) THEN ! east/west boundaries |
---|
1676 | DO jj = ijpt0, ijpt1 |
---|
1677 | DO ji = iipt0, iipt1 |
---|
1678 | ztab(ji,jj) = ptab(jj,jk) |
---|
1679 | END DO |
---|
1680 | END DO |
---|
1681 | ENDIF |
---|
1682 | |
---|
1683 | |
---|
1684 | ! 1. East and west directions |
---|
1685 | ! --------------------------- |
---|
1686 | ! |
---|
1687 | IF( nbondi /= 2 ) THEN ! Read Dirichlet lateral conditions |
---|
1688 | iihom = nlci-nreci |
---|
1689 | DO jl = 1, jpreci |
---|
1690 | t2ew(:,jl,1) = ztab(jpreci+jl,:) |
---|
1691 | t2we(:,jl,1) = ztab(iihom +jl,:) |
---|
1692 | END DO |
---|
1693 | ENDIF |
---|
1694 | ! |
---|
1695 | ! ! Migrations |
---|
1696 | imigr=jpreci*jpj |
---|
1697 | ! |
---|
1698 | IF( nbondi == -1 ) THEN |
---|
1699 | CALL mppsend( 2, t2we(1,1,1), imigr, noea, ml_req1 ) |
---|
1700 | CALL mpprecv( 1, t2ew(1,1,2), imigr, noea ) |
---|
1701 | IF(l_isend) CALL mpi_wait( ml_req1, ml_stat, ml_err ) |
---|
1702 | ELSEIF( nbondi == 0 ) THEN |
---|
1703 | CALL mppsend( 1, t2ew(1,1,1), imigr, nowe, ml_req1 ) |
---|
1704 | CALL mppsend( 2, t2we(1,1,1), imigr, noea, ml_req2 ) |
---|
1705 | CALL mpprecv( 1, t2ew(1,1,2), imigr, noea ) |
---|
1706 | CALL mpprecv( 2, t2we(1,1,2), imigr, nowe ) |
---|
1707 | IF(l_isend) CALL mpi_wait( ml_req1, ml_stat, ml_err ) |
---|
1708 | IF(l_isend) CALL mpi_wait( ml_req2, ml_stat, ml_err ) |
---|
1709 | ELSEIF( nbondi == 1 ) THEN |
---|
1710 | CALL mppsend( 1, t2ew(1,1,1), imigr, nowe, ml_req1 ) |
---|
1711 | CALL mpprecv( 2, t2we(1,1,2), imigr, nowe ) |
---|
1712 | IF(l_isend) CALL mpi_wait( ml_req1, ml_stat, ml_err ) |
---|
1713 | ENDIF |
---|
1714 | ! |
---|
1715 | ! ! Write Dirichlet lateral conditions |
---|
1716 | iihom = nlci-jpreci |
---|
1717 | ! |
---|
1718 | IF( nbondi == 0 .OR. nbondi == 1 ) THEN |
---|
1719 | DO jl = 1, jpreci |
---|
1720 | ztab(jl,:) = t2we(:,jl,2) |
---|
1721 | END DO |
---|
1722 | ENDIF |
---|
1723 | IF( nbondi == -1 .OR. nbondi == 0 ) THEN |
---|
1724 | DO jl = 1, jpreci |
---|
1725 | ztab(iihom+jl,:) = t2ew(:,jl,2) |
---|
1726 | END DO |
---|
1727 | ENDIF |
---|
1728 | |
---|
1729 | |
---|
1730 | ! 2. North and south directions |
---|
1731 | ! ----------------------------- |
---|
1732 | ! |
---|
1733 | IF( nbondj /= 2 ) THEN ! Read Dirichlet lateral conditions |
---|
1734 | ijhom = nlcj-nrecj |
---|
1735 | DO jl = 1, jprecj |
---|
1736 | t2sn(:,jl,1) = ztab(:,ijhom +jl) |
---|
1737 | t2ns(:,jl,1) = ztab(:,jprecj+jl) |
---|
1738 | END DO |
---|
1739 | ENDIF |
---|
1740 | ! |
---|
1741 | ! ! Migrations |
---|
1742 | imigr = jprecj * jpi |
---|
1743 | ! |
---|
1744 | IF( nbondj == -1 ) THEN |
---|
1745 | CALL mppsend( 4, t2sn(1,1,1), imigr, nono, ml_req1 ) |
---|
1746 | CALL mpprecv( 3, t2ns(1,1,2), imigr, nono ) |
---|
1747 | IF(l_isend) CALL mpi_wait( ml_req1, ml_stat, ml_err ) |
---|
1748 | ELSEIF( nbondj == 0 ) THEN |
---|
1749 | CALL mppsend( 3, t2ns(1,1,1), imigr, noso, ml_req1 ) |
---|
1750 | CALL mppsend( 4, t2sn(1,1,1), imigr, nono, ml_req2 ) |
---|
1751 | CALL mpprecv( 3, t2ns(1,1,2), imigr, nono ) |
---|
1752 | CALL mpprecv( 4, t2sn(1,1,2), imigr, noso ) |
---|
1753 | IF( l_isend ) CALL mpi_wait( ml_req1, ml_stat, ml_err ) |
---|
1754 | IF( l_isend ) CALL mpi_wait( ml_req2, ml_stat, ml_err ) |
---|
1755 | ELSEIF( nbondj == 1 ) THEN |
---|
1756 | CALL mppsend( 3, t2ns(1,1,1), imigr, noso, ml_req1 ) |
---|
1757 | CALL mpprecv( 4, t2sn(1,1,2), imigr, noso) |
---|
1758 | IF( l_isend ) CALL mpi_wait( ml_req1, ml_stat, ml_err ) |
---|
1759 | ENDIF |
---|
1760 | ! |
---|
1761 | ! ! Write Dirichlet lateral conditions |
---|
1762 | ijhom = nlcj - jprecj |
---|
1763 | IF( nbondj == 0 .OR. nbondj == 1 ) THEN |
---|
1764 | DO jl = 1, jprecj |
---|
1765 | ztab(:,jl) = t2sn(:,jl,2) |
---|
1766 | END DO |
---|
1767 | ENDIF |
---|
1768 | IF( nbondj == 0 .OR. nbondj == -1 ) THEN |
---|
1769 | DO jl = 1, jprecj |
---|
1770 | ztab(:,ijhom+jl) = t2ns(:,jl,2) |
---|
1771 | END DO |
---|
1772 | ENDIF |
---|
1773 | IF( ktype==1 .AND. kd1 <= jpi+nimpp-1 .AND. nimpp <= kd2 ) THEN |
---|
1774 | DO jj = ijpt0, ijpt1 ! north/south boundaries |
---|
1775 | DO ji = iipt0,ilpt1 |
---|
1776 | ptab(ji,jk) = ztab(ji,jj) |
---|
1777 | END DO |
---|
1778 | END DO |
---|
1779 | ELSEIF( ktype==2 .AND. kd1 <= jpj+njmpp-1 .AND. njmpp <= kd2 ) THEN |
---|
1780 | DO jj = ijpt0, ilpt1 ! east/west boundaries |
---|
1781 | DO ji = iipt0,iipt1 |
---|
1782 | ptab(jj,jk) = ztab(ji,jj) |
---|
1783 | END DO |
---|
1784 | END DO |
---|
1785 | ENDIF |
---|
1786 | ! |
---|
1787 | END DO |
---|
1788 | ! |
---|
1789 | END SUBROUTINE mppobc |
---|
1790 | |
---|
1791 | |
---|
1792 | SUBROUTINE mpp_comm_free( kcom ) |
---|
1793 | !!---------------------------------------------------------------------- |
---|
1794 | !!---------------------------------------------------------------------- |
---|
1795 | INTEGER, INTENT(in) :: kcom |
---|
1796 | !! |
---|
1797 | INTEGER :: ierr |
---|
1798 | !!---------------------------------------------------------------------- |
---|
1799 | ! |
---|
1800 | CALL MPI_COMM_FREE(kcom, ierr) |
---|
1801 | ! |
---|
1802 | END SUBROUTINE mpp_comm_free |
---|
1803 | |
---|
1804 | |
---|
1805 | SUBROUTINE mpp_ini_ice( pindic ) |
---|
1806 | !!---------------------------------------------------------------------- |
---|
1807 | !! *** routine mpp_ini_ice *** |
---|
1808 | !! |
---|
1809 | !! ** Purpose : Initialize special communicator for ice areas |
---|
1810 | !! condition together with global variables needed in the ddmpp folding |
---|
1811 | !! |
---|
1812 | !! ** Method : - Look for ice processors in ice routines |
---|
1813 | !! - Put their number in nrank_ice |
---|
1814 | !! - Create groups for the world processors and the ice processors |
---|
1815 | !! - Create a communicator for ice processors |
---|
1816 | !! |
---|
1817 | !! ** output |
---|
1818 | !! njmppmax = njmpp for northern procs |
---|
1819 | !! ndim_rank_ice = number of processors with ice |
---|
1820 | !! nrank_ice (ndim_rank_ice) = ice processors |
---|
1821 | !! ngrp_world = group ID for the world processors |
---|
1822 | !! ngrp_ice = group ID for the ice processors |
---|
1823 | !! ncomm_ice = communicator for the ice procs. |
---|
1824 | !! n_ice_root = number (in the world) of proc 0 in the ice comm. |
---|
1825 | !! |
---|
1826 | !!---------------------------------------------------------------------- |
---|
1827 | INTEGER, INTENT(in) :: pindic |
---|
1828 | !! |
---|
1829 | INTEGER :: ierr |
---|
1830 | INTEGER :: jjproc |
---|
1831 | INTEGER :: ii |
---|
1832 | INTEGER, DIMENSION(jpnij) :: kice |
---|
1833 | INTEGER, DIMENSION(jpnij) :: zwork |
---|
1834 | !!---------------------------------------------------------------------- |
---|
1835 | ! |
---|
1836 | ! Look for how many procs with sea-ice |
---|
1837 | ! |
---|
1838 | kice = 0 |
---|
1839 | DO jjproc = 1, jpnij |
---|
1840 | IF( jjproc == narea .AND. pindic .GT. 0 ) kice(jjproc) = 1 |
---|
1841 | END DO |
---|
1842 | ! |
---|
1843 | zwork = 0 |
---|
1844 | CALL MPI_ALLREDUCE( kice, zwork, jpnij, mpi_integer, mpi_sum, mpi_comm_opa, ierr ) |
---|
1845 | ndim_rank_ice = SUM( zwork ) |
---|
1846 | |
---|
1847 | ! Allocate the right size to nrank_north |
---|
1848 | #if ! defined key_agrif |
---|
1849 | IF( ALLOCATED ( nrank_ice ) ) DEALLOCATE( nrank_ice ) |
---|
1850 | #else |
---|
1851 | IF( ASSOCIATED( nrank_ice ) ) DEALLOCATE( nrank_ice ) |
---|
1852 | #endif |
---|
1853 | ALLOCATE( nrank_ice(ndim_rank_ice) ) |
---|
1854 | ! |
---|
1855 | ii = 0 |
---|
1856 | nrank_ice = 0 |
---|
1857 | DO jjproc = 1, jpnij |
---|
1858 | IF( zwork(jjproc) == 1) THEN |
---|
1859 | ii = ii + 1 |
---|
1860 | nrank_ice(ii) = jjproc -1 |
---|
1861 | ENDIF |
---|
1862 | END DO |
---|
1863 | |
---|
1864 | ! Create the world group |
---|
1865 | CALL MPI_COMM_GROUP( mpi_comm_opa, ngrp_world, ierr ) |
---|
1866 | |
---|
1867 | ! Create the ice group from the world group |
---|
1868 | CALL MPI_GROUP_INCL( ngrp_world, ndim_rank_ice, nrank_ice, ngrp_ice, ierr ) |
---|
1869 | |
---|
1870 | ! Create the ice communicator , ie the pool of procs with sea-ice |
---|
1871 | CALL MPI_COMM_CREATE( mpi_comm_opa, ngrp_ice, ncomm_ice, ierr ) |
---|
1872 | |
---|
1873 | ! Find proc number in the world of proc 0 in the north |
---|
1874 | ! The following line seems to be useless, we just comment & keep it as reminder |
---|
1875 | ! CALL MPI_GROUP_TRANSLATE_RANKS(ngrp_ice,1,0,ngrp_world,n_ice_root,ierr) |
---|
1876 | ! |
---|
1877 | END SUBROUTINE mpp_ini_ice |
---|
1878 | |
---|
1879 | |
---|
1880 | SUBROUTINE mpp_ini_znl |
---|
1881 | !!---------------------------------------------------------------------- |
---|
1882 | !! *** routine mpp_ini_znl *** |
---|
1883 | !! |
---|
1884 | !! ** Purpose : Initialize special communicator for computing zonal sum |
---|
1885 | !! |
---|
1886 | !! ** Method : - Look for processors in the same row |
---|
1887 | !! - Put their number in nrank_znl |
---|
1888 | !! - Create group for the znl processors |
---|
1889 | !! - Create a communicator for znl processors |
---|
1890 | !! - Determine if processor should write znl files |
---|
1891 | !! |
---|
1892 | !! ** output |
---|
1893 | !! ndim_rank_znl = number of processors on the same row |
---|
1894 | !! ngrp_znl = group ID for the znl processors |
---|
1895 | !! ncomm_znl = communicator for the ice procs. |
---|
1896 | !! n_znl_root = number (in the world) of proc 0 in the ice comm. |
---|
1897 | !! |
---|
1898 | !!---------------------------------------------------------------------- |
---|
1899 | INTEGER :: ierr |
---|
1900 | INTEGER :: jproc |
---|
1901 | INTEGER :: ii |
---|
1902 | INTEGER, DIMENSION(jpnij) :: kwork |
---|
1903 | ! |
---|
1904 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - ngrp_world : ', ngrp_world |
---|
1905 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - mpi_comm_world : ', mpi_comm_world |
---|
1906 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - mpi_comm_opa : ', mpi_comm_opa |
---|
1907 | ! |
---|
1908 | IF ( jpnj == 1 ) THEN |
---|
1909 | ngrp_znl = ngrp_world |
---|
1910 | ncomm_znl = mpi_comm_opa |
---|
1911 | ELSE |
---|
1912 | ! |
---|
1913 | CALL MPI_ALLGATHER ( njmpp, 1, mpi_integer, kwork, 1, mpi_integer, mpi_comm_opa, ierr ) |
---|
1914 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - kwork pour njmpp : ', kwork |
---|
1915 | !-$$ CALL flush(numout) |
---|
1916 | ! |
---|
1917 | ! Count number of processors on the same row |
---|
1918 | ndim_rank_znl = 0 |
---|
1919 | DO jproc=1,jpnij |
---|
1920 | IF ( kwork(jproc) == njmpp ) THEN |
---|
1921 | ndim_rank_znl = ndim_rank_znl + 1 |
---|
1922 | ENDIF |
---|
1923 | END DO |
---|
1924 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - ndim_rank_znl : ', ndim_rank_znl |
---|
1925 | !-$$ CALL flush(numout) |
---|
1926 | ! Allocate the right size to nrank_znl |
---|
1927 | #if ! defined key_agrif |
---|
1928 | IF (ALLOCATED (nrank_znl)) DEALLOCATE(nrank_znl) |
---|
1929 | #else |
---|
1930 | IF (ASSOCIATED(nrank_znl)) DEALLOCATE(nrank_znl) |
---|
1931 | #endif |
---|
1932 | ALLOCATE(nrank_znl(ndim_rank_znl)) |
---|
1933 | ii = 0 |
---|
1934 | nrank_znl (:) = 0 |
---|
1935 | DO jproc=1,jpnij |
---|
1936 | IF ( kwork(jproc) == njmpp) THEN |
---|
1937 | ii = ii + 1 |
---|
1938 | nrank_znl(ii) = jproc -1 |
---|
1939 | ENDIF |
---|
1940 | END DO |
---|
1941 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - nrank_znl : ', nrank_znl |
---|
1942 | !-$$ CALL flush(numout) |
---|
1943 | |
---|
1944 | ! Create the opa group |
---|
1945 | CALL MPI_COMM_GROUP(mpi_comm_opa,ngrp_opa,ierr) |
---|
1946 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - ngrp_opa : ', ngrp_opa |
---|
1947 | !-$$ CALL flush(numout) |
---|
1948 | |
---|
1949 | ! Create the znl group from the opa group |
---|
1950 | CALL MPI_GROUP_INCL ( ngrp_opa, ndim_rank_znl, nrank_znl, ngrp_znl, ierr ) |
---|
1951 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - ngrp_znl ', ngrp_znl |
---|
1952 | !-$$ CALL flush(numout) |
---|
1953 | |
---|
1954 | ! Create the znl communicator from the opa communicator, ie the pool of procs in the same row |
---|
1955 | CALL MPI_COMM_CREATE ( mpi_comm_opa, ngrp_znl, ncomm_znl, ierr ) |
---|
1956 | !-$$ WRITE (numout,*) 'mpp_ini_znl ', nproc, ' - ncomm_znl ', ncomm_znl |
---|
1957 | !-$$ CALL flush(numout) |
---|
1958 | ! |
---|
1959 | END IF |
---|
1960 | |
---|
1961 | ! Determines if processor if the first (starting from i=1) on the row |
---|
1962 | IF ( jpni == 1 ) THEN |
---|
1963 | l_znl_root = .TRUE. |
---|
1964 | ELSE |
---|
1965 | l_znl_root = .FALSE. |
---|
1966 | kwork (1) = nimpp |
---|
1967 | CALL mpp_min ( kwork(1), kcom = ncomm_znl) |
---|
1968 | IF ( nimpp == kwork(1)) l_znl_root = .TRUE. |
---|
1969 | END IF |
---|
1970 | |
---|
1971 | END SUBROUTINE mpp_ini_znl |
---|
1972 | |
---|
1973 | |
---|
1974 | SUBROUTINE mpp_ini_north |
---|
1975 | !!---------------------------------------------------------------------- |
---|
1976 | !! *** routine mpp_ini_north *** |
---|
1977 | !! |
---|
1978 | !! ** Purpose : Initialize special communicator for north folding |
---|
1979 | !! condition together with global variables needed in the mpp folding |
---|
1980 | !! |
---|
1981 | !! ** Method : - Look for northern processors |
---|
1982 | !! - Put their number in nrank_north |
---|
1983 | !! - Create groups for the world processors and the north processors |
---|
1984 | !! - Create a communicator for northern processors |
---|
1985 | !! |
---|
1986 | !! ** output |
---|
1987 | !! njmppmax = njmpp for northern procs |
---|
1988 | !! ndim_rank_north = number of processors in the northern line |
---|
1989 | !! nrank_north (ndim_rank_north) = number of the northern procs. |
---|
1990 | !! ngrp_world = group ID for the world processors |
---|
1991 | !! ngrp_north = group ID for the northern processors |
---|
1992 | !! ncomm_north = communicator for the northern procs. |
---|
1993 | !! north_root = number (in the world) of proc 0 in the northern comm. |
---|
1994 | !! |
---|
1995 | !!---------------------------------------------------------------------- |
---|
1996 | INTEGER :: ierr |
---|
1997 | INTEGER :: jjproc |
---|
1998 | INTEGER :: ii, ji |
---|
1999 | !!---------------------------------------------------------------------- |
---|
2000 | ! |
---|
2001 | njmppmax = MAXVAL( njmppt ) |
---|
2002 | ! |
---|
2003 | ! Look for how many procs on the northern boundary |
---|
2004 | ndim_rank_north = 0 |
---|
2005 | DO jjproc = 1, jpnij |
---|
2006 | IF( njmppt(jjproc) == njmppmax ) ndim_rank_north = ndim_rank_north + 1 |
---|
2007 | END DO |
---|
2008 | ! |
---|
2009 | ! Allocate the right size to nrank_north |
---|
2010 | #if ! defined key_agrif |
---|
2011 | IF (ALLOCATED (nrank_north)) DEALLOCATE(nrank_north) |
---|
2012 | #else |
---|
2013 | IF (ASSOCIATED(nrank_north)) DEALLOCATE(nrank_north) |
---|
2014 | #endif |
---|
2015 | ALLOCATE( nrank_north(ndim_rank_north) ) |
---|
2016 | |
---|
2017 | ! Fill the nrank_north array with proc. number of northern procs. |
---|
2018 | ! Note : the rank start at 0 in MPI |
---|
2019 | ii = 0 |
---|
2020 | DO ji = 1, jpnij |
---|
2021 | IF ( njmppt(ji) == njmppmax ) THEN |
---|
2022 | ii=ii+1 |
---|
2023 | nrank_north(ii)=ji-1 |
---|
2024 | END IF |
---|
2025 | END DO |
---|
2026 | ! |
---|
2027 | ! create the world group |
---|
2028 | CALL MPI_COMM_GROUP( mpi_comm_opa, ngrp_world, ierr ) |
---|
2029 | ! |
---|
2030 | ! Create the North group from the world group |
---|
2031 | CALL MPI_GROUP_INCL( ngrp_world, ndim_rank_north, nrank_north, ngrp_north, ierr ) |
---|
2032 | ! |
---|
2033 | ! Create the North communicator , ie the pool of procs in the north group |
---|
2034 | CALL MPI_COMM_CREATE( mpi_comm_opa, ngrp_north, ncomm_north, ierr ) |
---|
2035 | ! |
---|
2036 | END SUBROUTINE mpp_ini_north |
---|
2037 | |
---|
2038 | |
---|
2039 | SUBROUTINE mpp_lbc_north_3d( pt3d, cd_type, psgn ) |
---|
2040 | !!--------------------------------------------------------------------- |
---|
2041 | !! *** routine mpp_lbc_north_3d *** |
---|
2042 | !! |
---|
2043 | !! ** Purpose : Ensure proper north fold horizontal bondary condition |
---|
2044 | !! in mpp configuration in case of jpn1 > 1 |
---|
2045 | !! |
---|
2046 | !! ** Method : North fold condition and mpp with more than one proc |
---|
2047 | !! in i-direction require a specific treatment. We gather |
---|
2048 | !! the 4 northern lines of the global domain on 1 processor |
---|
2049 | !! and apply lbc north-fold on this sub array. Then we |
---|
2050 | !! scatter the north fold array back to the processors. |
---|
2051 | !! |
---|
2052 | !!---------------------------------------------------------------------- |
---|
2053 | REAL(wp), DIMENSION(jpi,jpj,jpk), INTENT(inout) :: pt3d ! 3D array on which the b.c. is applied |
---|
2054 | CHARACTER(len=1) , INTENT(in ) :: cd_type ! nature of pt3d grid-points |
---|
2055 | ! ! = T , U , V , F or W gridpoints |
---|
2056 | REAL(wp) , INTENT(in ) :: psgn ! = -1. the sign change across the north fold |
---|
2057 | !! ! = 1. , the sign is kept |
---|
2058 | INTEGER :: ji, jj, jr |
---|
2059 | INTEGER :: ierr, itaille, ildi, ilei, iilb |
---|
2060 | INTEGER :: ijpj, ijpjm1, ij, iproc |
---|
2061 | REAL(wp), DIMENSION(jpiglo,4,jpk) :: ztab |
---|
2062 | REAL(wp), DIMENSION(jpi ,4,jpk) :: znorthloc |
---|
2063 | REAL(wp), DIMENSION(jpi ,4,jpk,jpni) :: znorthgloio |
---|
2064 | REAL(wp), DIMENSION(jpi, 4,jpk) :: zfoldwrk ! Workspace for message transfers avoiding mpi_allgather |
---|
2065 | INTEGER, DIMENSION (jpmaxngh) :: ml_req5 ! for mpi_isend when avoiding mpi_allgather |
---|
2066 | INTEGER :: ml_err ! for mpi_isend when avoiding mpi_allgather |
---|
2067 | INTEGER, DIMENSION(MPI_STATUS_SIZE) :: ml_stat ! for mpi_isend when avoiding mpi_allgather |
---|
2068 | !!---------------------------------------------------------------------- |
---|
2069 | ! |
---|
2070 | ijpj = 4 |
---|
2071 | ityp = -1 |
---|
2072 | ijpjm1 = 3 |
---|
2073 | ztab(:,:,:) = 0.e0 |
---|
2074 | ! |
---|
2075 | DO jj = nlcj - ijpj +1, nlcj ! put in znorthloc the last 4 jlines of pt3d |
---|
2076 | ij = jj - nlcj + ijpj |
---|
2077 | znorthloc(:,ij,:) = pt3d(:,jj,:) |
---|
2078 | END DO |
---|
2079 | ! |
---|
2080 | ! ! Build in procs of ncomm_north the znorthgloio |
---|
2081 | itaille = jpi * jpk * ijpj |
---|
2082 | IF ( lnorth_nogather ) THEN |
---|
2083 | ! |
---|
2084 | ! Avoid the use of mpi_allgather by exchanging only with the processes already identified (in opa_northcomms) |
---|
2085 | ! as being involved in this process' northern boundary exchange |
---|
2086 | ! |
---|
2087 | ! First put local values into the global arraay |
---|
2088 | DO jj = nlcj-ijpj+1, nlcj |
---|
2089 | ij = jj - nlcj + ijpj |
---|
2090 | DO ji = 1, nlci |
---|
2091 | ztab(ji+nimpp-1,ij,:) = pt3d(ji,jj,:) |
---|
2092 | END DO |
---|
2093 | END DO |
---|
2094 | |
---|
2095 | ! |
---|
2096 | ! Set the exchange type in order to access the correct list of active neighbours |
---|
2097 | ! |
---|
2098 | SELECT CASE ( cd_type ) |
---|
2099 | CASE ( 'T' , 'W' ) |
---|
2100 | ityp = 1 |
---|
2101 | CASE ( 'U' ) |
---|
2102 | ityp = 2 |
---|
2103 | CASE ( 'V' ) |
---|
2104 | ityp = 3 |
---|
2105 | CASE ( 'F' ) |
---|
2106 | ityp = 4 |
---|
2107 | CASE DEFAULT |
---|
2108 | ! |
---|
2109 | ! Set a default value for unsupported types which will cause a fallback to |
---|
2110 | ! the mpi_allgather method |
---|
2111 | ! |
---|
2112 | ityp = -1 |
---|
2113 | END SELECT |
---|
2114 | IF ( ityp .gt. 0 ) THEN |
---|
2115 | |
---|
2116 | DO jr = 1,nsndto(ityp) |
---|
2117 | CALL mppsend(5, znorthloc, itaille, isendto(jr,ityp), ml_req5(jr) ) |
---|
2118 | END DO |
---|
2119 | DO jr = 1,nsndto(ityp) |
---|
2120 | CALL mpprecv(5, zfoldwrk, itaille, isendto(jr,ityp)) |
---|
2121 | iproc = isendto(jr,ityp) + 1 |
---|
2122 | ildi=nldit (iproc) |
---|
2123 | ilei=nleit (iproc) |
---|
2124 | iilb=nimppt(iproc) |
---|
2125 | DO jj = 1, 4 |
---|
2126 | DO ji = ildi, ilei |
---|
2127 | ztab(ji+iilb-1,jj,:) = zfoldwrk(ji,jj,:) |
---|
2128 | END DO |
---|
2129 | END DO |
---|
2130 | END DO |
---|
2131 | IF(l_isend) THEN |
---|
2132 | DO jr = 1,nsndto(ityp) |
---|
2133 | CALL mpi_wait(ml_req5(jr), ml_stat, ml_err) |
---|
2134 | END DO |
---|
2135 | ENDIF |
---|
2136 | |
---|
2137 | ENDIF |
---|
2138 | |
---|
2139 | ENDIF |
---|
2140 | |
---|
2141 | IF ( ityp .lt. 0 ) THEN |
---|
2142 | CALL MPI_ALLGATHER( znorthloc , itaille, MPI_DOUBLE_PRECISION, & |
---|
2143 | & znorthgloio, itaille, MPI_DOUBLE_PRECISION, ncomm_north, ierr ) |
---|
2144 | ! |
---|
2145 | ! ! recover the global north array |
---|
2146 | DO jr = 1, ndim_rank_north |
---|
2147 | iproc = nrank_north(jr) + 1 |
---|
2148 | ildi = nldit (iproc) |
---|
2149 | ilei = nleit (iproc) |
---|
2150 | iilb = nimppt(iproc) |
---|
2151 | DO jj = 1, 4 |
---|
2152 | DO ji = ildi, ilei |
---|
2153 | ztab(ji+iilb-1,jj,:) = znorthgloio(ji,jj,:,jr) |
---|
2154 | END DO |
---|
2155 | END DO |
---|
2156 | END DO |
---|
2157 | ENDIF |
---|
2158 | ! |
---|
2159 | ! The ztab array has been either: |
---|
2160 | ! a. Fully populated by the mpi_allgather operation or |
---|
2161 | ! b. Had the active points for this domain and northern neighbours populated by peer to peer exchanges |
---|
2162 | ! Either way the array may be folded by lbc_nfd and the result for the span of this domain will be identical. |
---|
2163 | ! |
---|
2164 | CALL lbc_nfd( ztab, cd_type, psgn ) ! North fold boundary condition |
---|
2165 | ! |
---|
2166 | DO jj = nlcj-ijpj+1, nlcj ! Scatter back to pt3d |
---|
2167 | ij = jj - nlcj + ijpj |
---|
2168 | DO ji= 1, nlci |
---|
2169 | pt3d(ji,jj,:) = ztab(ji+nimpp-1,ij,:) |
---|
2170 | END DO |
---|
2171 | END DO |
---|
2172 | ! |
---|
2173 | END SUBROUTINE mpp_lbc_north_3d |
---|
2174 | |
---|
2175 | |
---|
2176 | SUBROUTINE mpp_lbc_north_2d( pt2d, cd_type, psgn) |
---|
2177 | !!--------------------------------------------------------------------- |
---|
2178 | !! *** routine mpp_lbc_north_2d *** |
---|
2179 | !! |
---|
2180 | !! ** Purpose : Ensure proper north fold horizontal bondary condition |
---|
2181 | !! in mpp configuration in case of jpn1 > 1 (for 2d array ) |
---|
2182 | !! |
---|
2183 | !! ** Method : North fold condition and mpp with more than one proc |
---|
2184 | !! in i-direction require a specific treatment. We gather |
---|
2185 | !! the 4 northern lines of the global domain on 1 processor |
---|
2186 | !! and apply lbc north-fold on this sub array. Then we |
---|
2187 | !! scatter the north fold array back to the processors. |
---|
2188 | !! |
---|
2189 | !!---------------------------------------------------------------------- |
---|
2190 | REAL(wp), DIMENSION(jpi,jpj), INTENT(inout) :: pt2d ! 3D array on which the b.c. is applied |
---|
2191 | CHARACTER(len=1) , INTENT(in ) :: cd_type ! nature of pt3d grid-points |
---|
2192 | ! ! = T , U , V , F or W gridpoints |
---|
2193 | REAL(wp) , INTENT(in ) :: psgn ! = -1. the sign change across the north fold |
---|
2194 | !! ! = 1. , the sign is kept |
---|
2195 | INTEGER :: ji, jj, jr |
---|
2196 | INTEGER :: ierr, itaille, ildi, ilei, iilb |
---|
2197 | INTEGER :: ijpj, ijpjm1, ij, iproc |
---|
2198 | REAL(wp), DIMENSION(jpiglo,4) :: ztab |
---|
2199 | REAL(wp), DIMENSION(jpi ,4) :: znorthloc |
---|
2200 | REAL(wp), DIMENSION(jpi ,4,jpni) :: znorthgloio |
---|
2201 | REAL(wp), DIMENSION(jpi, 4) :: zfoldwrk ! Workspace for message transfers avoiding mpi_allgather |
---|
2202 | INTEGER, DIMENSION (jpmaxngh) :: ml_req5 ! for mpi_isend when avoiding mpi_allgather |
---|
2203 | INTEGER :: ml_err ! for mpi_isend when avoiding mpi_allgather |
---|
2204 | INTEGER, DIMENSION(MPI_STATUS_SIZE):: ml_stat ! for mpi_isend when avoiding mpi_allgather |
---|
2205 | !!---------------------------------------------------------------------- |
---|
2206 | ! |
---|
2207 | ijpj = 4 |
---|
2208 | ityp = -1 |
---|
2209 | ijpjm1 = 3 |
---|
2210 | ztab(:,:) = 0.e0 |
---|
2211 | ! |
---|
2212 | DO jj = nlcj-ijpj+1, nlcj ! put in znorthloc the last 4 jlines of pt2d |
---|
2213 | ij = jj - nlcj + ijpj |
---|
2214 | znorthloc(:,ij) = pt2d(:,jj) |
---|
2215 | END DO |
---|
2216 | |
---|
2217 | ! ! Build in procs of ncomm_north the znorthgloio |
---|
2218 | itaille = jpi * ijpj |
---|
2219 | IF ( lnorth_nogather ) THEN |
---|
2220 | ! |
---|
2221 | ! Avoid the use of mpi_allgather by exchanging only with the processes already identified (in opa_northcomms) |
---|
2222 | ! as being involved in this process' northern boundary exchange |
---|
2223 | ! |
---|
2224 | ! First put local values into the global array |
---|
2225 | ! |
---|
2226 | DO jj = nlcj-ijpj+1, nlcj |
---|
2227 | ij = jj - nlcj + ijpj |
---|
2228 | DO ji = 1, nlci |
---|
2229 | ztab(ji+nimpp-1,ij) = pt2d(ji,jj) |
---|
2230 | END DO |
---|
2231 | END DO |
---|
2232 | |
---|
2233 | ! |
---|
2234 | ! Set the exchange type in order to access the correct list of active neighbours |
---|
2235 | ! |
---|
2236 | SELECT CASE ( cd_type ) |
---|
2237 | CASE ( 'T' , 'W' ) |
---|
2238 | ityp = 1 |
---|
2239 | CASE ( 'U' ) |
---|
2240 | ityp = 2 |
---|
2241 | CASE ( 'V' ) |
---|
2242 | ityp = 3 |
---|
2243 | CASE ( 'F' ) |
---|
2244 | ityp = 4 |
---|
2245 | CASE DEFAULT |
---|
2246 | ! |
---|
2247 | ! Set a default value for unsupported types which will cause a fallback to |
---|
2248 | ! the mpi_allgather method |
---|
2249 | ! |
---|
2250 | ityp = -1 |
---|
2251 | END SELECT |
---|
2252 | |
---|
2253 | IF ( ityp .gt. 0 ) THEN |
---|
2254 | |
---|
2255 | DO jr = 1,nsndto(ityp) |
---|
2256 | CALL mppsend(5, znorthloc, itaille, isendto(jr,ityp), ml_req5(jr) ) |
---|
2257 | END DO |
---|
2258 | DO jr = 1,nsndto(ityp) |
---|
2259 | CALL mpprecv(5, zfoldwrk, itaille, isendto(jr,ityp)) |
---|
2260 | iproc = isendto(jr,ityp) + 1 |
---|
2261 | ildi=nldit (iproc) |
---|
2262 | ilei=nleit (iproc) |
---|
2263 | iilb=nimppt(iproc) |
---|
2264 | DO jj = 1, 4 |
---|
2265 | DO ji = ildi, ilei |
---|
2266 | ztab(ji+iilb-1,jj) = zfoldwrk(ji,jj) |
---|
2267 | END DO |
---|
2268 | END DO |
---|
2269 | END DO |
---|
2270 | IF(l_isend) THEN |
---|
2271 | DO jr = 1,nsndto(ityp) |
---|
2272 | CALL mpi_wait(ml_req5(jr), ml_stat, ml_err) |
---|
2273 | END DO |
---|
2274 | ENDIF |
---|
2275 | |
---|
2276 | ENDIF |
---|
2277 | |
---|
2278 | ENDIF |
---|
2279 | |
---|
2280 | IF ( ityp .lt. 0 ) THEN |
---|
2281 | CALL MPI_ALLGATHER( znorthloc , itaille, MPI_DOUBLE_PRECISION, & |
---|
2282 | & znorthgloio, itaille, MPI_DOUBLE_PRECISION, ncomm_north, ierr ) |
---|
2283 | ! |
---|
2284 | DO jr = 1, ndim_rank_north ! recover the global north array |
---|
2285 | iproc = nrank_north(jr) + 1 |
---|
2286 | ildi=nldit (iproc) |
---|
2287 | ilei=nleit (iproc) |
---|
2288 | iilb=nimppt(iproc) |
---|
2289 | DO jj = 1, 4 |
---|
2290 | DO ji = ildi, ilei |
---|
2291 | ztab(ji+iilb-1,jj) = znorthgloio(ji,jj,jr) |
---|
2292 | END DO |
---|
2293 | END DO |
---|
2294 | END DO |
---|
2295 | ENDIF |
---|
2296 | ! |
---|
2297 | ! The ztab array has been either: |
---|
2298 | ! a. Fully populated by the mpi_allgather operation or |
---|
2299 | ! b. Had the active points for this domain and northern neighbours populated by peer to peer exchanges |
---|
2300 | ! Either way the array may be folded by lbc_nfd and the result for the span of this domain will be identical. |
---|
2301 | ! |
---|
2302 | CALL lbc_nfd( ztab, cd_type, psgn ) ! North fold boundary condition |
---|
2303 | ! |
---|
2304 | ! |
---|
2305 | DO jj = nlcj-ijpj+1, nlcj ! Scatter back to pt2d |
---|
2306 | ij = jj - nlcj + ijpj |
---|
2307 | DO ji = 1, nlci |
---|
2308 | pt2d(ji,jj) = ztab(ji+nimpp-1,ij) |
---|
2309 | END DO |
---|
2310 | END DO |
---|
2311 | ! |
---|
2312 | END SUBROUTINE mpp_lbc_north_2d |
---|
2313 | |
---|
2314 | |
---|
2315 | SUBROUTINE mpp_lbc_north_e( pt2d, cd_type, psgn) |
---|
2316 | !!--------------------------------------------------------------------- |
---|
2317 | !! *** routine mpp_lbc_north_2d *** |
---|
2318 | !! |
---|
2319 | !! ** Purpose : Ensure proper north fold horizontal bondary condition |
---|
2320 | !! in mpp configuration in case of jpn1 > 1 and for 2d |
---|
2321 | !! array with outer extra halo |
---|
2322 | !! |
---|
2323 | !! ** Method : North fold condition and mpp with more than one proc |
---|
2324 | !! in i-direction require a specific treatment. We gather |
---|
2325 | !! the 4+2*jpr2dj northern lines of the global domain on 1 |
---|
2326 | !! processor and apply lbc north-fold on this sub array. |
---|
2327 | !! Then we scatter the north fold array back to the processors. |
---|
2328 | !! |
---|
2329 | !!---------------------------------------------------------------------- |
---|
2330 | REAL(wp), DIMENSION(1-jpr2di:jpi+jpr2di,1-jpr2dj:jpj+jpr2dj), INTENT(inout) :: pt2d ! 2D array with extra halo |
---|
2331 | CHARACTER(len=1) , INTENT(in ) :: cd_type ! nature of pt3d grid-points |
---|
2332 | ! ! = T , U , V , F or W -points |
---|
2333 | REAL(wp) , INTENT(in ) :: psgn ! = -1. the sign change across the |
---|
2334 | !! ! north fold, = 1. otherwise |
---|
2335 | INTEGER :: ji, jj, jr |
---|
2336 | INTEGER :: ierr, itaille, ildi, ilei, iilb |
---|
2337 | INTEGER :: ijpj, ij, iproc |
---|
2338 | REAL(wp), DIMENSION(jpiglo,4+2*jpr2dj) :: ztab |
---|
2339 | REAL(wp), DIMENSION(jpi ,4+2*jpr2dj) :: znorthloc |
---|
2340 | REAL(wp), DIMENSION(jpi ,4+2*jpr2dj,jpni) :: znorthgloio |
---|
2341 | !!---------------------------------------------------------------------- |
---|
2342 | ! |
---|
2343 | ijpj=4 |
---|
2344 | ztab(:,:) = 0.e0 |
---|
2345 | |
---|
2346 | ij=0 |
---|
2347 | ! put in znorthloc the last 4 jlines of pt2d |
---|
2348 | DO jj = nlcj - ijpj + 1 - jpr2dj, nlcj +jpr2dj |
---|
2349 | ij = ij + 1 |
---|
2350 | DO ji = 1, jpi |
---|
2351 | znorthloc(ji,ij)=pt2d(ji,jj) |
---|
2352 | END DO |
---|
2353 | END DO |
---|
2354 | ! |
---|
2355 | itaille = jpi * ( ijpj + 2 * jpr2dj ) |
---|
2356 | CALL MPI_ALLGATHER( znorthloc(1,1) , itaille, MPI_DOUBLE_PRECISION, & |
---|
2357 | & znorthgloio(1,1,1), itaille, MPI_DOUBLE_PRECISION, ncomm_north, ierr ) |
---|
2358 | ! |
---|
2359 | DO jr = 1, ndim_rank_north ! recover the global north array |
---|
2360 | iproc = nrank_north(jr) + 1 |
---|
2361 | ildi = nldit (iproc) |
---|
2362 | ilei = nleit (iproc) |
---|
2363 | iilb = nimppt(iproc) |
---|
2364 | DO jj = 1, ijpj+2*jpr2dj |
---|
2365 | DO ji = ildi, ilei |
---|
2366 | ztab(ji+iilb-1,jj) = znorthgloio(ji,jj,jr) |
---|
2367 | END DO |
---|
2368 | END DO |
---|
2369 | END DO |
---|
2370 | |
---|
2371 | |
---|
2372 | ! 2. North-Fold boundary conditions |
---|
2373 | ! ---------------------------------- |
---|
2374 | CALL lbc_nfd( ztab(:,:), cd_type, psgn, pr2dj = jpr2dj ) |
---|
2375 | |
---|
2376 | ij = jpr2dj |
---|
2377 | !! Scatter back to pt2d |
---|
2378 | DO jj = nlcj - ijpj + 1 , nlcj +jpr2dj |
---|
2379 | ij = ij +1 |
---|
2380 | DO ji= 1, nlci |
---|
2381 | pt2d(ji,jj) = ztab(ji+nimpp-1,ij) |
---|
2382 | END DO |
---|
2383 | END DO |
---|
2384 | ! |
---|
2385 | END SUBROUTINE mpp_lbc_north_e |
---|
2386 | |
---|
2387 | |
---|
2388 | SUBROUTINE mpi_init_opa( code ) |
---|
2389 | !!--------------------------------------------------------------------- |
---|
2390 | !! *** routine mpp_init.opa *** |
---|
2391 | !! |
---|
2392 | !! ** Purpose :: export and attach a MPI buffer for bsend |
---|
2393 | !! |
---|
2394 | !! ** Method :: define buffer size in namelist, if 0 no buffer attachment |
---|
2395 | !! but classical mpi_init |
---|
2396 | !! |
---|
2397 | !! History :: 01/11 :: IDRIS initial version for IBM only |
---|
2398 | !! 08/04 :: R. Benshila, generalisation |
---|
2399 | !!--------------------------------------------------------------------- |
---|
2400 | INTEGER :: code, ierr |
---|
2401 | LOGICAL :: mpi_was_called |
---|
2402 | !!--------------------------------------------------------------------- |
---|
2403 | ! |
---|
2404 | CALL mpi_initialized( mpi_was_called, code ) ! MPI initialization |
---|
2405 | IF ( code /= MPI_SUCCESS ) THEN |
---|
2406 | CALL ctl_stop( ' lib_mpp: Error in routine mpi_initialized' ) |
---|
2407 | CALL mpi_abort( mpi_comm_world, code, ierr ) |
---|
2408 | ENDIF |
---|
2409 | ! |
---|
2410 | IF( .NOT. mpi_was_called ) THEN |
---|
2411 | CALL mpi_init( code ) |
---|
2412 | CALL mpi_comm_dup( mpi_comm_world, mpi_comm_opa, code ) |
---|
2413 | IF ( code /= MPI_SUCCESS ) THEN |
---|
2414 | CALL ctl_stop( ' lib_mpp: Error in routine mpi_comm_dup' ) |
---|
2415 | CALL mpi_abort( mpi_comm_world, code, ierr ) |
---|
2416 | ENDIF |
---|
2417 | ENDIF |
---|
2418 | ! |
---|
2419 | IF( nn_buffer > 0 ) THEN |
---|
2420 | IF ( lwp ) WRITE(numout,*) 'mpi_bsend, buffer allocation of : ', nn_buffer |
---|
2421 | ! Buffer allocation and attachment |
---|
2422 | ALLOCATE( tampon(nn_buffer) ) |
---|
2423 | CALL mpi_buffer_attach( tampon, nn_buffer,code ) |
---|
2424 | ENDIF |
---|
2425 | ! |
---|
2426 | END SUBROUTINE mpi_init_opa |
---|
2427 | |
---|
2428 | #else |
---|
2429 | !!---------------------------------------------------------------------- |
---|
2430 | !! Default case: Dummy module share memory computing |
---|
2431 | !!---------------------------------------------------------------------- |
---|
2432 | INTERFACE mpp_sum |
---|
2433 | MODULE PROCEDURE mpp_sum_a2s, mpp_sum_as, mpp_sum_ai, mpp_sum_s, mpp_sum_i |
---|
2434 | END INTERFACE |
---|
2435 | INTERFACE mpp_max |
---|
2436 | MODULE PROCEDURE mppmax_a_int, mppmax_int, mppmax_a_real, mppmax_real |
---|
2437 | END INTERFACE |
---|
2438 | INTERFACE mpp_min |
---|
2439 | MODULE PROCEDURE mppmin_a_int, mppmin_int, mppmin_a_real, mppmin_real |
---|
2440 | END INTERFACE |
---|
2441 | INTERFACE mppobc |
---|
2442 | MODULE PROCEDURE mppobc_1d, mppobc_2d, mppobc_3d, mppobc_4d |
---|
2443 | END INTERFACE |
---|
2444 | INTERFACE mpp_minloc |
---|
2445 | MODULE PROCEDURE mpp_minloc2d ,mpp_minloc3d |
---|
2446 | END INTERFACE |
---|
2447 | INTERFACE mpp_maxloc |
---|
2448 | MODULE PROCEDURE mpp_maxloc2d ,mpp_maxloc3d |
---|
2449 | END INTERFACE |
---|
2450 | |
---|
2451 | |
---|
2452 | LOGICAL, PUBLIC, PARAMETER :: lk_mpp = .FALSE. !: mpp flag |
---|
2453 | INTEGER :: ncomm_ice |
---|
2454 | |
---|
2455 | CONTAINS |
---|
2456 | |
---|
2457 | FUNCTION mynode( ldtxt, localComm ) RESULT (function_value) |
---|
2458 | CHARACTER(len=*),DIMENSION(:), INTENT( out) :: ldtxt |
---|
2459 | INTEGER, OPTIONAL , INTENT(in ) :: localComm |
---|
2460 | IF( PRESENT( localComm ) .OR. .NOT.PRESENT( localComm ) ) function_value = 0 |
---|
2461 | IF( .FALSE. ) ldtxt(:) = 'never done' |
---|
2462 | END FUNCTION mynode |
---|
2463 | |
---|
2464 | SUBROUTINE mppsync ! Dummy routine |
---|
2465 | END SUBROUTINE mppsync |
---|
2466 | |
---|
2467 | SUBROUTINE mpp_sum_as( parr, kdim, kcom ) ! Dummy routine |
---|
2468 | REAL , DIMENSION(:) :: parr |
---|
2469 | INTEGER :: kdim |
---|
2470 | INTEGER, OPTIONAL :: kcom |
---|
2471 | WRITE(*,*) 'mpp_sum_as: You should not have seen this print! error?', kdim, parr(1), kcom |
---|
2472 | END SUBROUTINE mpp_sum_as |
---|
2473 | |
---|
2474 | SUBROUTINE mpp_sum_a2s( parr, kdim, kcom ) ! Dummy routine |
---|
2475 | REAL , DIMENSION(:,:) :: parr |
---|
2476 | INTEGER :: kdim |
---|
2477 | INTEGER, OPTIONAL :: kcom |
---|
2478 | WRITE(*,*) 'mpp_sum_a2s: You should not have seen this print! error?', kdim, parr(1,1), kcom |
---|
2479 | END SUBROUTINE mpp_sum_a2s |
---|
2480 | |
---|
2481 | SUBROUTINE mpp_sum_ai( karr, kdim, kcom ) ! Dummy routine |
---|
2482 | INTEGER, DIMENSION(:) :: karr |
---|
2483 | INTEGER :: kdim |
---|
2484 | INTEGER, OPTIONAL :: kcom |
---|
2485 | WRITE(*,*) 'mpp_sum_ai: You should not have seen this print! error?', kdim, karr(1), kcom |
---|
2486 | END SUBROUTINE mpp_sum_ai |
---|
2487 | |
---|
2488 | SUBROUTINE mpp_sum_s( psca, kcom ) ! Dummy routine |
---|
2489 | REAL :: psca |
---|
2490 | INTEGER, OPTIONAL :: kcom |
---|
2491 | WRITE(*,*) 'mpp_sum_s: You should not have seen this print! error?', psca, kcom |
---|
2492 | END SUBROUTINE mpp_sum_s |
---|
2493 | |
---|
2494 | SUBROUTINE mpp_sum_i( kint, kcom ) ! Dummy routine |
---|
2495 | integer :: kint |
---|
2496 | INTEGER, OPTIONAL :: kcom |
---|
2497 | WRITE(*,*) 'mpp_sum_i: You should not have seen this print! error?', kint, kcom |
---|
2498 | END SUBROUTINE mpp_sum_i |
---|
2499 | |
---|
2500 | SUBROUTINE mppmax_a_real( parr, kdim, kcom ) |
---|
2501 | REAL , DIMENSION(:) :: parr |
---|
2502 | INTEGER :: kdim |
---|
2503 | INTEGER, OPTIONAL :: kcom |
---|
2504 | WRITE(*,*) 'mppmax_a_real: You should not have seen this print! error?', kdim, parr(1), kcom |
---|
2505 | END SUBROUTINE mppmax_a_real |
---|
2506 | |
---|
2507 | SUBROUTINE mppmax_real( psca, kcom ) |
---|
2508 | REAL :: psca |
---|
2509 | INTEGER, OPTIONAL :: kcom |
---|
2510 | WRITE(*,*) 'mppmax_real: You should not have seen this print! error?', psca, kcom |
---|
2511 | END SUBROUTINE mppmax_real |
---|
2512 | |
---|
2513 | SUBROUTINE mppmin_a_real( parr, kdim, kcom ) |
---|
2514 | REAL , DIMENSION(:) :: parr |
---|
2515 | INTEGER :: kdim |
---|
2516 | INTEGER, OPTIONAL :: kcom |
---|
2517 | WRITE(*,*) 'mppmin_a_real: You should not have seen this print! error?', kdim, parr(1), kcom |
---|
2518 | END SUBROUTINE mppmin_a_real |
---|
2519 | |
---|
2520 | SUBROUTINE mppmin_real( psca, kcom ) |
---|
2521 | REAL :: psca |
---|
2522 | INTEGER, OPTIONAL :: kcom |
---|
2523 | WRITE(*,*) 'mppmin_real: You should not have seen this print! error?', psca, kcom |
---|
2524 | END SUBROUTINE mppmin_real |
---|
2525 | |
---|
2526 | SUBROUTINE mppmax_a_int( karr, kdim ,kcom) |
---|
2527 | INTEGER, DIMENSION(:) :: karr |
---|
2528 | INTEGER :: kdim |
---|
2529 | INTEGER, OPTIONAL :: kcom |
---|
2530 | WRITE(*,*) 'mppmax_a_int: You should not have seen this print! error?', kdim, karr(1), kcom |
---|
2531 | END SUBROUTINE mppmax_a_int |
---|
2532 | |
---|
2533 | SUBROUTINE mppmax_int( kint, kcom) |
---|
2534 | INTEGER :: kint |
---|
2535 | INTEGER, OPTIONAL :: kcom |
---|
2536 | WRITE(*,*) 'mppmax_int: You should not have seen this print! error?', kint, kcom |
---|
2537 | END SUBROUTINE mppmax_int |
---|
2538 | |
---|
2539 | SUBROUTINE mppmin_a_int( karr, kdim, kcom ) |
---|
2540 | INTEGER, DIMENSION(:) :: karr |
---|
2541 | INTEGER :: kdim |
---|
2542 | INTEGER, OPTIONAL :: kcom |
---|
2543 | WRITE(*,*) 'mppmin_a_int: You should not have seen this print! error?', kdim, karr(1), kcom |
---|
2544 | END SUBROUTINE mppmin_a_int |
---|
2545 | |
---|
2546 | SUBROUTINE mppmin_int( kint, kcom ) |
---|
2547 | INTEGER :: kint |
---|
2548 | INTEGER, OPTIONAL :: kcom |
---|
2549 | WRITE(*,*) 'mppmin_int: You should not have seen this print! error?', kint, kcom |
---|
2550 | END SUBROUTINE mppmin_int |
---|
2551 | |
---|
2552 | SUBROUTINE mppobc_1d( parr, kd1, kd2, kl, kk, ktype, kij ) |
---|
2553 | INTEGER :: kd1, kd2, kl , kk, ktype, kij |
---|
2554 | REAL, DIMENSION(:) :: parr ! variable array |
---|
2555 | WRITE(*,*) 'mppobc: You should not have seen this print! error?', parr(1), kd1, kd2, kl, kk, ktype, kij |
---|
2556 | END SUBROUTINE mppobc_1d |
---|
2557 | |
---|
2558 | SUBROUTINE mppobc_2d( parr, kd1, kd2, kl, kk, ktype, kij ) |
---|
2559 | INTEGER :: kd1, kd2, kl , kk, ktype, kij |
---|
2560 | REAL, DIMENSION(:,:) :: parr ! variable array |
---|
2561 | WRITE(*,*) 'mppobc: You should not have seen this print! error?', parr(1,1), kd1, kd2, kl, kk, ktype, kij |
---|
2562 | END SUBROUTINE mppobc_2d |
---|
2563 | |
---|
2564 | SUBROUTINE mppobc_3d( parr, kd1, kd2, kl, kk, ktype, kij ) |
---|
2565 | INTEGER :: kd1, kd2, kl , kk, ktype, kij |
---|
2566 | REAL, DIMENSION(:,:,:) :: parr ! variable array |
---|
2567 | WRITE(*,*) 'mppobc: You should not have seen this print! error?', parr(1,1,1), kd1, kd2, kl, kk, ktype, kij |
---|
2568 | END SUBROUTINE mppobc_3d |
---|
2569 | |
---|
2570 | SUBROUTINE mppobc_4d( parr, kd1, kd2, kl, kk, ktype, kij ) |
---|
2571 | INTEGER :: kd1, kd2, kl , kk, ktype, kij |
---|
2572 | REAL, DIMENSION(:,:,:,:) :: parr ! variable array |
---|
2573 | WRITE(*,*) 'mppobc: You should not have seen this print! error?', parr(1,1,1,1), kd1, kd2, kl, kk, ktype, kij |
---|
2574 | END SUBROUTINE mppobc_4d |
---|
2575 | |
---|
2576 | SUBROUTINE mpp_minloc2d( ptab, pmask, pmin, ki, kj ) |
---|
2577 | REAL :: pmin |
---|
2578 | REAL , DIMENSION (:,:) :: ptab, pmask |
---|
2579 | INTEGER :: ki, kj |
---|
2580 | WRITE(*,*) 'mpp_minloc2d: You should not have seen this print! error?', pmin, ki, kj, ptab(1,1), pmask(1,1) |
---|
2581 | END SUBROUTINE mpp_minloc2d |
---|
2582 | |
---|
2583 | SUBROUTINE mpp_minloc3d( ptab, pmask, pmin, ki, kj, kk ) |
---|
2584 | REAL :: pmin |
---|
2585 | REAL , DIMENSION (:,:,:) :: ptab, pmask |
---|
2586 | INTEGER :: ki, kj, kk |
---|
2587 | WRITE(*,*) 'mpp_minloc3d: You should not have seen this print! error?', pmin, ki, kj, kk, ptab(1,1,1), pmask(1,1,1) |
---|
2588 | END SUBROUTINE mpp_minloc3d |
---|
2589 | |
---|
2590 | SUBROUTINE mpp_maxloc2d( ptab, pmask, pmax, ki, kj ) |
---|
2591 | REAL :: pmax |
---|
2592 | REAL , DIMENSION (:,:) :: ptab, pmask |
---|
2593 | INTEGER :: ki, kj |
---|
2594 | WRITE(*,*) 'mpp_maxloc2d: You should not have seen this print! error?', pmax, ki, kj, ptab(1,1), pmask(1,1) |
---|
2595 | END SUBROUTINE mpp_maxloc2d |
---|
2596 | |
---|
2597 | SUBROUTINE mpp_maxloc3d( ptab, pmask, pmax, ki, kj, kk ) |
---|
2598 | REAL :: pmax |
---|
2599 | REAL , DIMENSION (:,:,:) :: ptab, pmask |
---|
2600 | INTEGER :: ki, kj, kk |
---|
2601 | WRITE(*,*) 'mpp_maxloc3d: You should not have seen this print! error?', pmax, ki, kj, kk, ptab(1,1,1), pmask(1,1,1) |
---|
2602 | END SUBROUTINE mpp_maxloc3d |
---|
2603 | |
---|
2604 | SUBROUTINE mppstop |
---|
2605 | WRITE(*,*) 'mppstop: You should not have seen this print! error?' |
---|
2606 | END SUBROUTINE mppstop |
---|
2607 | |
---|
2608 | SUBROUTINE mpp_ini_ice( kcom ) |
---|
2609 | INTEGER :: kcom |
---|
2610 | WRITE(*,*) 'mpp_ini_ice: You should not have seen this print! error?', kcom |
---|
2611 | END SUBROUTINE mpp_ini_ice |
---|
2612 | |
---|
2613 | SUBROUTINE mpp_ini_znl |
---|
2614 | WRITE(*,*) 'mpp_ini_znl: You should not have seen this print! error?' |
---|
2615 | END SUBROUTINE mpp_ini_znl |
---|
2616 | |
---|
2617 | SUBROUTINE mpp_comm_free( kcom ) |
---|
2618 | INTEGER :: kcom |
---|
2619 | WRITE(*,*) 'mpp_comm_free: You should not have seen this print! error?', kcom |
---|
2620 | END SUBROUTINE mpp_comm_free |
---|
2621 | |
---|
2622 | #endif |
---|
2623 | !!---------------------------------------------------------------------- |
---|
2624 | END MODULE lib_mpp |
---|