source: palm/trunk/SOURCE/palm.f90 @ 1779

Last change on this file since 1779 was 1779, checked in by raasch, 5 years ago

pmc array management changed from linked list to sequential loop; further small changes and cosmetics for the pmc

  • Property svn:keywords set to Id
File size: 14.9 KB
RevLine 
[1682]1!> @file palm.f90
[1036]2!--------------------------------------------------------------------------------!
3! This file is part of PALM.
4!
5! PALM is free software: you can redistribute it and/or modify it under the terms
6! of the GNU General Public License as published by the Free Software Foundation,
7! either version 3 of the License, or (at your option) any later version.
8!
9! PALM is distributed in the hope that it will be useful, but WITHOUT ANY
10! WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
11! A PARTICULAR PURPOSE.  See the GNU General Public License for more details.
12!
13! You should have received a copy of the GNU General Public License along with
14! PALM. If not, see <http://www.gnu.org/licenses/>.
15!
[1310]16! Copyright 1997-2014 Leibniz Universitaet Hannover
[1036]17!--------------------------------------------------------------------------------!
18!
[484]19! Current revisions:
[1]20! -----------------
[1779]21! setting of nest_domain and coupling_char moved to the pmci
[1748]22!
[1321]23! Former revisions:
24! -----------------
25! $Id: palm.f90 1779 2016-03-03 08:01:28Z raasch $
26!
[1765]27! 1764 2016-02-28 12:45:19Z raasch
28! cpp-statements for nesting removed, communicator settings cleaned up
29!
[1763]30! 1762 2016-02-25 12:31:13Z hellstea
31! Introduction of nested domain feature
32!
[1748]33! 1747 2016-02-08 12:25:53Z raasch
34! OpenACC-adjustment for new surface layer parameterization
35!
[1683]36! 1682 2015-10-07 23:56:08Z knoop
37! Code annotations made doxygen readable
38!
[1669]39! 1668 2015-09-23 13:45:36Z raasch
40! warning replaced by abort in case of failed user interface check
41!
[1667]42! 1666 2015-09-23 07:31:10Z raasch
43! check for user's interface version added
44!
[1483]45! 1482 2014-10-18 12:34:45Z raasch
46! adjustments for using CUDA-aware OpenMPI
47!
[1469]48! 1468 2014-09-24 14:06:57Z maronga
49! Adapted for use on up to 6-digit processor cores
50!
[1403]51! 1402 2014-05-09 14:25:13Z raasch
52! location messages added
53!
[1375]54! 1374 2014-04-25 12:55:07Z raasch
55! bugfix: various modules added
56!
[1321]57! 1320 2014-03-20 08:40:49Z raasch
[1320]58! ONLY-attribute added to USE-statements,
59! kind-parameters added to all INTEGER and REAL declaration statements,
60! kinds are defined in new module kinds,
61! old module precision_kind is removed,
62! revision history before 2012 removed,
63! comment fields (!:) to be used for variable explanations added to
64! all variable declaration statements
[77]65!
[1319]66! 1318 2014-03-17 13:35:16Z raasch
67! module interfaces removed
68!
[1242]69! 1241 2013-10-30 11:36:58Z heinze
70! initialization of nuding and large scale forcing from external file
71!
[1222]72! 1221 2013-09-10 08:59:13Z raasch
73! +wall_flags_00, rflags_invers, rflags_s_inner in copyin statement
74!
[1213]75! 1212 2013-08-15 08:46:27Z raasch
76! +tri in copyin statement
77!
[1182]78! 1179 2013-06-14 05:57:58Z raasch
79! ref_state added to copyin-list
80!
[1114]81! 1113 2013-03-10 02:48:14Z raasch
82! openACC statements modified
83!
[1112]84! 1111 2013-03-08 23:54:10Z raasch
85! openACC statements updated
86!
[1093]87! 1092 2013-02-02 11:24:22Z raasch
88! unused variables removed
89!
[1037]90! 1036 2012-10-22 13:43:42Z raasch
91! code put under GPL (PALM 3.9)
92!
[1017]93! 1015 2012-09-27 09:23:24Z raasch
[863]94! Version number changed from 3.8 to 3.8a.
[1017]95! OpenACC statements added + code changes required for GPU optimization
[863]96!
[850]97! 849 2012-03-15 10:35:09Z raasch
98! write_particles renamed lpm_write_restart_file
99!
[1]100! Revision 1.1  1997/07/24 11:23:35  raasch
101! Initial revision
102!
103!
104! Description:
105! ------------
[1682]106!> Large-Eddy Simulation (LES) model for the convective boundary layer,
107!> optimized for use on parallel machines (implementation realized using the
108!> Message Passing Interface (MPI)). The model can also be run on vector machines
109!> (less well optimized) and workstations. Versions for the different types of
110!> machines are controlled via cpp-directives.
111!> Model runs are only feasible using the ksh-script mrun.
[1]112!------------------------------------------------------------------------------!
[1682]113 PROGRAM palm
114 
[1]115
[1374]116    USE arrays_3d
[1]117
[1320]118    USE control_parameters,                                                    &
119        ONLY:  coupling_char, coupling_mode, do2d_at_begin, do3d_at_begin,     &
[1666]120               io_blocks, io_group, large_scale_forcing, message_string,       &
[1762]121               nest_domain, nudging, simulated_time, simulated_time_chr,       &
[1666]122               user_interface_current_revision,                                &
123               user_interface_required_revision, version, wall_heatflux,       &
124               write_binary
[1320]125
126    USE cpulog,                                                                &
127        ONLY:  cpu_log, log_point, cpu_statistics
128
[1374]129    USE grid_variables,                                                        &
130        ONLY:  fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, wall_e_x, wall_e_y, &
131               wall_u, wall_v, wall_w_x, wall_w_y
132
133    USE indices,                                                               &
[1666]134        ONLY:  ngp_2dh, ngp_2dh_s_inner, nzb_diff_s_inner, nzb_diff_s_outer,   &
135               nzb_diff_u, nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner,  &
136               nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner,             &
137               nzb_w_outer, rflags_invers, rflags_s_inner, wall_flags_0,       &
138               wall_flags_00
[1374]139
[1320]140    USE kinds
141
142    USE ls_forcing_mod,                                                        &
143        ONLY:  init_ls_forcing
144
145    USE nudge_mod,                                                             &
146        ONLY:  init_nudge
147
148    USE particle_attributes,                                                   &
149        ONLY:  particle_advection
150
[1]151    USE pegrid
152
[1762]153    USE pmc_interface,                                                         &
[1764]154        ONLY:  cpl_id, nested_run, pmci_init, pmci_modelconfiguration
[1762]155
[1374]156    USE statistics,                                                            &
157        ONLY:  hom, rmask, weight_pres, weight_substep
158
[1747]159    USE surface_layer_fluxes_mod,                                              &
160        ONLY:  pt1, qv1, uv_total
161
[1015]162#if defined( __openacc )
163    USE OPENACC
164#endif
165
[1]166    IMPLICIT NONE
167
168!
169!-- Local variables
[1682]170    CHARACTER(LEN=9)  ::  time_to_string  !<
171    CHARACTER(LEN=10) ::  env_string      !< to store string of environment var
172    INTEGER(iwp)      ::  env_stat        !< to hold status of GET_ENV
173    INTEGER(iwp)      ::  i               !<
174    INTEGER(iwp)      ::  myid_openmpi    !< OpenMPI local rank for CUDA aware MPI
[1015]175#if defined( __openacc )
[1682]176    REAL(wp), DIMENSION(100) ::  acc_dum     !<
[1015]177#endif
[1]178
[1539]179    version = 'PALM 4.0'
[1666]180    user_interface_required_revision = 'r1663'
[75]181
[1]182#if defined( __parallel )
183!
184!-- MPI initialisation. comm2d is preliminary set, because
185!-- it will be defined in init_pegrid but is used before in cpu_log.
186    CALL MPI_INIT( ierr )
[1762]187
188!
189!-- Initialize the coupling for nested-domain runs
[1764]190!-- comm_palm is the communicator which includes all PEs (MPI processes)
191!-- available for this (nested) model. If it is not a nested run, comm_palm
192!-- is returned as MPI_COMM_WORLD
[1762]193    CALL pmci_init( comm_palm )
194    comm2d = comm_palm
[1764]195!
196!-- Get the (preliminary) number of MPI processes and the local PE-id (in case
197!-- of a further communicator splitting in init_coupling, these numbers will
198!-- be changed in init_pegrid).
199    IF ( nested_run )  THEN
[1762]200
[1764]201       CALL MPI_COMM_SIZE( comm_palm, numprocs, ierr )
202       CALL MPI_COMM_RANK( comm_palm, myid, ierr )
[1762]203
[1764]204    ELSE
205
206       CALL MPI_COMM_SIZE( MPI_COMM_WORLD, numprocs, ierr )
207       CALL MPI_COMM_RANK( MPI_COMM_WORLD, myid, ierr )
[1]208!
[1764]209!--    Initialize PE topology in case of coupled atmosphere-ocean runs (comm_palm
210!--    will be splitted in init_coupling)
211       CALL init_coupling
212    ENDIF
[102]213#endif
214
[1015]215#if defined( __openacc )
[102]216!
[1482]217!-- Get the local MPI rank in case of CUDA aware OpenMPI. Important, if there
218!-- is more than one accelerator board on the node
219    CALL GET_ENVIRONMENT_VARIABLE('OMPI_COMM_WORLD_LOCAL_RANK',                &
220         VALUE=env_string, STATUS=env_stat )
221    READ( env_string, '(I1)' )  myid_openmpi
222    PRINT*, '### local_rank = ', myid_openmpi, '  status=',env_stat
223!
[1015]224!-- Get the number of accelerator boards per node and assign the MPI processes
225!-- to these boards
[1092]226    PRINT*, '*** ACC_DEVICE_NVIDIA = ', ACC_DEVICE_NVIDIA
[1015]227    num_acc_per_node  = ACC_GET_NUM_DEVICES( ACC_DEVICE_NVIDIA )
[1092]228    IF ( numprocs == 1  .AND.  num_acc_per_node > 0 )  num_acc_per_node = 1
[1482]229    PRINT*, '*** myid = ', myid_openmpi, ' num_acc_per_node = ', num_acc_per_node
230    acc_rank = MOD( myid_openmpi, num_acc_per_node )
[1015]231    CALL ACC_SET_DEVICE_NUM ( acc_rank, ACC_DEVICE_NVIDIA )
232!
233!-- Test output (to be removed later)
[1482]234    WRITE (*,'(A,I6,A,I3,A,I3,A,I3)') '*** Connect MPI-Task ', myid_openmpi,   &
235                                      ' to CPU ', acc_rank, ' Devices: ',      &
236                                      num_acc_per_node, ' connected to:',      &
[1015]237                                      ACC_GET_DEVICE_NUM( ACC_DEVICE_NVIDIA )
238#endif
[1092]239
[1015]240!
241!-- Ensure that OpenACC first attaches the GPU devices by copying a dummy data
242!-- region
243    !$acc data copyin( acc_dum )
244
245!
[1]246!-- Initialize measuring of the CPU-time remaining to the run
247    CALL local_tremain_ini
248
249!
250!-- Start of total CPU time measuring.
251    CALL cpu_log( log_point(1), 'total', 'start' )
252    CALL cpu_log( log_point(2), 'initialisation', 'start' )
253
254!
[206]255!-- Open a file for debug output
[1468]256    WRITE (myid_char,'(''_'',I6.6)')  myid
[206]257    OPEN( 9, FILE='DEBUG'//TRIM( coupling_char )//myid_char, FORM='FORMATTED' )
258
259!
[1]260!-- Initialize dvrp logging. Also, one PE maybe split from the global
261!-- communicator for doing the dvrp output. In that case, the number of
262!-- PEs available for PALM is reduced by one and communicator comm_palm
263!-- is changed respectively.
264#if defined( __parallel )
265    CALL MPI_COMM_RANK( comm_palm, myid, ierr )
266!
[102]267!-- TEST OUTPUT (TO BE REMOVED)
268    WRITE(9,*) '*** coupling_mode = "', TRIM( coupling_mode ), '"'
269    CALL LOCAL_FLUSH( 9 )
[215]270    IF ( TRIM( coupling_mode ) /= 'uncoupled' )  THEN
271       PRINT*, '*** PE', myid, ' Global target PE:', target_id, &
272               TRIM( coupling_mode )
273    ENDIF
[102]274#endif
275
[108]276    CALL init_dvrp_logging
277
[102]278!
[108]279!-- Read control parameters from NAMELIST files and read environment-variables
280    CALL parin
281
282!
[1666]283!-- Check for the user's interface version
284    IF ( user_interface_current_revision /= user_interface_required_revision )  &
285    THEN
286       message_string = 'current user-interface revision "' //                  &
287                        TRIM( user_interface_current_revision ) // '" does ' // &
288                        'not match the required revision ' //                   &
289                        TRIM( user_interface_required_revision )
[1668]290        CALL message( 'palm', 'PA0169', 1, 2, 0, 6, 0 )
[1666]291    ENDIF
292
293!
[108]294!-- Determine processor topology and local array indices
295    CALL init_pegrid
296
297!
[1]298!-- Generate grid parameters
299    CALL init_grid
300
301!
[1241]302!-- Initialize nudging if required
303    IF ( nudging )  THEN
304       CALL init_nudge
305    ENDIF
306
307!
308!-- Initialize reading of large scale forcing from external file - if required
309    IF ( large_scale_forcing )  THEN
310       CALL init_ls_forcing
311    ENDIF
312
313!
[1]314!-- Check control parameters and deduce further quantities
315    CALL check_parameters
316
317!
318!-- Initialize all necessary variables
319    CALL init_3d_model
320
321!
[1762]322!-- Coupling protocol setup for nested-domain runs
[1764]323    IF ( nested_run )  THEN
324       CALL pmci_modelconfiguration
325    ENDIF
[1762]326
327!
[1]328!-- Output of program header
329    IF ( myid == 0 )  CALL header
330
331    CALL cpu_log( log_point(2), 'initialisation', 'stop' )
332
333!
334!-- Set start time in format hh:mm:ss
335    simulated_time_chr = time_to_string( simulated_time )
336
337!
338!-- If required, output of initial arrays
339    IF ( do2d_at_begin )  THEN
340       CALL data_output_2d( 'xy', 0 )
341       CALL data_output_2d( 'xz', 0 )
342       CALL data_output_2d( 'yz', 0 )
343    ENDIF
344    IF ( do3d_at_begin )  THEN
345       CALL data_output_3d( 0 )
346    ENDIF
347
348!
[1015]349!-- Declare and initialize variables in the accelerator memory with their
350!-- host values
[1113]351    !$acc  data copyin( d, diss, e, e_p, kh, km, p, pt, pt_p, q, ql, tend, te_m, tpt_m, tu_m, tv_m, tw_m, u, u_p, v, vpt, v_p, w, w_p )          &
[1212]352    !$acc       copyin( tri, tric, dzu, ddzu, ddzw, dd2zu, l_grid, l_wall, ptdf_x, ptdf_y, pt_init, rdf, rdf_sc, ref_state, ug, u_init, vg, v_init, zu, zw )   &
[1747]353    !$acc       copyin( hom, ol, pt1, qs, qsws, qswst, qv1, rif_wall, shf, ts, tswst, us, usws, uswst, uv_total, vsws, vswst, z0, z0h )      &
[1015]354    !$acc       copyin( fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, nzb_diff_s_inner, nzb_diff_s_outer, nzb_diff_u )       &
355    !$acc       copyin( nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner )    &
356    !$acc       copyin( nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner )   &
[1221]357    !$acc       copyin( nzb_w_outer, rflags_invers, rflags_s_inner, rmask, wall_heatflux, wall_e_x, wall_e_y, wall_u, wall_v, wall_w_x, wall_w_y, wall_flags_0, wall_flags_00 )  &
358    !$acc       copyin( ngp_2dh, ngp_2dh_s_inner )  &
[1113]359    !$acc       copyin( weight_pres, weight_substep )
[1015]360!
[495]361!-- Integration of the model equations using timestep-scheme
[1]362    CALL time_integration
363
364!
[495]365!-- If required, write binary data for restart runs
366    IF ( write_binary(1:4) == 'true' )  THEN
[759]367
368       CALL cpu_log( log_point(22), 'write_3d_binary', 'start' )
369
[1402]370       CALL location_message( 'writing restart data', .FALSE. )
371
[759]372       CALL check_open( 14 )
373
374       DO  i = 0, io_blocks-1
375          IF ( i == io_group )  THEN
[1]376!
[759]377!--          Write flow field data
378             CALL write_3d_binary
379          ENDIF
380#if defined( __parallel )
381          CALL MPI_BARRIER( comm2d, ierr )
382#endif
383       ENDDO
384
[1402]385       CALL location_message( 'finished', .TRUE. )
386
[759]387       CALL cpu_log( log_point(22), 'write_3d_binary', 'stop' )
388
[495]389!
390!--    If required, write particle data
[849]391       IF ( particle_advection )  CALL lpm_write_restart_file
[495]392    ENDIF
[1]393
394!
395!-- If required, repeat output of header including the required CPU-time
396    IF ( myid == 0 )  CALL header
397!
398!-- If required, final user-defined actions, and
399!-- last actions on the open files and close files. Unit 14 was opened
400!-- in write_3d_binary but it is closed here, to allow writing on this
401!-- unit in routine user_last_actions.
402    CALL cpu_log( log_point(4), 'last actions', 'start' )
[759]403    DO  i = 0, io_blocks-1
404       IF ( i == io_group )  THEN
405          CALL user_last_actions
406          IF ( write_binary(1:4) == 'true' )  CALL close_file( 14 )
407       ENDIF
408#if defined( __parallel )
409       CALL MPI_BARRIER( comm2d, ierr )
410#endif
411    ENDDO
[1]412    CALL close_file( 0 )
413    CALL close_dvrp
414    CALL cpu_log( log_point(4), 'last actions', 'stop' )
415
[102]416#if defined( __mpi2 )
[1]417!
[206]418!-- Test exchange via intercommunicator in case of a MPI-2 coupling
[102]419    IF ( coupling_mode == 'atmosphere_to_ocean' )  THEN
420       i = 12345 + myid
421       CALL MPI_SEND( i, 1, MPI_INTEGER, myid, 11, comm_inter, ierr )
422    ELSEIF ( coupling_mode == 'ocean_to_atmosphere' )  THEN
423       CALL MPI_RECV( i, 1, MPI_INTEGER, myid, 11, comm_inter, status, ierr )
424       PRINT*, '### myid: ', myid, '   received from atmosphere:  i = ', i
425    ENDIF
426#endif
427
428!
[1015]429!-- Close the OpenACC dummy data region
430    !$acc end data
431    !$acc end data
432
433!
[1]434!-- Take final CPU-time for CPU-time analysis
435    CALL cpu_log( log_point(1), 'total', 'stop' )
436    CALL cpu_statistics
437
438#if defined( __parallel )
439    CALL MPI_FINALIZE( ierr )
440#endif
441
442 END PROGRAM palm
Note: See TracBrowser for help on using the repository browser.