source: palm/trunk/SOURCE/palm.f90 @ 1807

Last change on this file since 1807 was 1784, checked in by raasch, 9 years ago

last commit documented

  • Property svn:keywords set to Id
File size: 15.9 KB
Line 
1!> @file palm.f90
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!
16! Copyright 1997-2014 Leibniz Universitaet Hannover
17!--------------------------------------------------------------------------------!
18!
19! Current revisions:
20! -----------------
21!
22!
23! Former revisions:
24! -----------------
25! $Id: palm.f90 1784 2016-03-06 19:14:40Z gronemeier $
26!
27! 1783 2016-03-06 18:36:17Z raasch
28! required user interface version changed
29!
30! 1781 2016-03-03 15:12:23Z raasch
31! pmc initialization moved from time_integration to here
32!
33! 1779 2016-03-03 08:01:28Z raasch
34! setting of nest_domain and coupling_char moved to the pmci
35!
36! 1764 2016-02-28 12:45:19Z raasch
37! cpp-statements for nesting removed, communicator settings cleaned up
38!
39! 1762 2016-02-25 12:31:13Z hellstea
40! Introduction of nested domain feature
41!
42! 1747 2016-02-08 12:25:53Z raasch
43! OpenACC-adjustment for new surface layer parameterization
44!
45! 1682 2015-10-07 23:56:08Z knoop
46! Code annotations made doxygen readable
47!
48! 1668 2015-09-23 13:45:36Z raasch
49! warning replaced by abort in case of failed user interface check
50!
51! 1666 2015-09-23 07:31:10Z raasch
52! check for user's interface version added
53!
54! 1482 2014-10-18 12:34:45Z raasch
55! adjustments for using CUDA-aware OpenMPI
56!
57! 1468 2014-09-24 14:06:57Z maronga
58! Adapted for use on up to 6-digit processor cores
59!
60! 1402 2014-05-09 14:25:13Z raasch
61! location messages added
62!
63! 1374 2014-04-25 12:55:07Z raasch
64! bugfix: various modules added
65!
66! 1320 2014-03-20 08:40:49Z raasch
67! ONLY-attribute added to USE-statements,
68! kind-parameters added to all INTEGER and REAL declaration statements,
69! kinds are defined in new module kinds,
70! old module precision_kind is removed,
71! revision history before 2012 removed,
72! comment fields (!:) to be used for variable explanations added to
73! all variable declaration statements
74!
75! 1318 2014-03-17 13:35:16Z raasch
76! module interfaces removed
77!
78! 1241 2013-10-30 11:36:58Z heinze
79! initialization of nuding and large scale forcing from external file
80!
81! 1221 2013-09-10 08:59:13Z raasch
82! +wall_flags_00, rflags_invers, rflags_s_inner in copyin statement
83!
84! 1212 2013-08-15 08:46:27Z raasch
85! +tri in copyin statement
86!
87! 1179 2013-06-14 05:57:58Z raasch
88! ref_state added to copyin-list
89!
90! 1113 2013-03-10 02:48:14Z raasch
91! openACC statements modified
92!
93! 1111 2013-03-08 23:54:10Z raasch
94! openACC statements updated
95!
96! 1092 2013-02-02 11:24:22Z raasch
97! unused variables removed
98!
99! 1036 2012-10-22 13:43:42Z raasch
100! code put under GPL (PALM 3.9)
101!
102! 1015 2012-09-27 09:23:24Z raasch
103! Version number changed from 3.8 to 3.8a.
104! OpenACC statements added + code changes required for GPU optimization
105!
106! 849 2012-03-15 10:35:09Z raasch
107! write_particles renamed lpm_write_restart_file
108!
109! Revision 1.1  1997/07/24 11:23:35  raasch
110! Initial revision
111!
112!
113! Description:
114! ------------
115!> Large-Eddy Simulation (LES) model for the convective boundary layer,
116!> optimized for use on parallel machines (implementation realized using the
117!> Message Passing Interface (MPI)). The model can also be run on vector machines
118!> (less well optimized) and workstations. Versions for the different types of
119!> machines are controlled via cpp-directives.
120!> Model runs are only feasible using the ksh-script mrun.
121!------------------------------------------------------------------------------!
122 PROGRAM palm
123 
124
125    USE arrays_3d
126
127    USE control_parameters,                                                    &
128        ONLY:  constant_diffusion, coupling_char, coupling_mode,               &
129               do2d_at_begin, do3d_at_begin, humidity, io_blocks, io_group,    &
130               large_scale_forcing, message_string, nest_domain, nudging,      &
131               passive_scalar, simulated_time, simulated_time_chr,             &
132               user_interface_current_revision,                                &
133               user_interface_required_revision, version, wall_heatflux,       &
134               write_binary
135
136    USE cpulog,                                                                &
137        ONLY:  cpu_log, log_point, cpu_statistics
138
139    USE grid_variables,                                                        &
140        ONLY:  fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, wall_e_x, wall_e_y, &
141               wall_u, wall_v, wall_w_x, wall_w_y
142
143    USE indices,                                                               &
144        ONLY:  nbgp, ngp_2dh, ngp_2dh_s_inner, nzb_diff_s_inner, nzb_diff_s_outer,   &
145               nzb_diff_u, nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner,  &
146               nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner,             &
147               nzb_w_outer, rflags_invers, rflags_s_inner, wall_flags_0,       &
148               wall_flags_00
149
150    USE kinds
151
152    USE ls_forcing_mod,                                                        &
153        ONLY:  init_ls_forcing
154
155    USE nudge_mod,                                                             &
156        ONLY:  init_nudge
157
158    USE particle_attributes,                                                   &
159        ONLY:  particle_advection
160
161    USE pegrid
162
163    USE pmc_interface,                                                         &
164        ONLY:  cpl_id, nested_run, pmci_client_initialize, pmci_init,          &
165               pmci_modelconfiguration, pmci_server_initialize
166
167    USE statistics,                                                            &
168        ONLY:  hom, rmask, weight_pres, weight_substep
169
170    USE surface_layer_fluxes_mod,                                              &
171        ONLY:  pt1, qv1, uv_total
172
173#if defined( __openacc )
174    USE OPENACC
175#endif
176
177    IMPLICIT NONE
178
179!
180!-- Local variables
181    CHARACTER(LEN=9)  ::  time_to_string  !<
182    CHARACTER(LEN=10) ::  env_string      !< to store string of environment var
183    INTEGER(iwp)      ::  env_stat        !< to hold status of GET_ENV
184    INTEGER(iwp)      ::  i               !<
185    INTEGER(iwp)      ::  myid_openmpi    !< OpenMPI local rank for CUDA aware MPI
186#if defined( __openacc )
187    REAL(wp), DIMENSION(100) ::  acc_dum     !<
188#endif
189
190    version = 'PALM 4.0'
191    user_interface_required_revision = 'r1783'
192
193#if defined( __parallel )
194!
195!-- MPI initialisation. comm2d is preliminary set, because
196!-- it will be defined in init_pegrid but is used before in cpu_log.
197    CALL MPI_INIT( ierr )
198
199!
200!-- Initialize the coupling for nested-domain runs
201!-- comm_palm is the communicator which includes all PEs (MPI processes)
202!-- available for this (nested) model. If it is not a nested run, comm_palm
203!-- is returned as MPI_COMM_WORLD
204    CALL pmci_init( comm_palm )
205    comm2d = comm_palm
206!
207!-- Get the (preliminary) number of MPI processes and the local PE-id (in case
208!-- of a further communicator splitting in init_coupling, these numbers will
209!-- be changed in init_pegrid).
210    IF ( nested_run )  THEN
211
212       CALL MPI_COMM_SIZE( comm_palm, numprocs, ierr )
213       CALL MPI_COMM_RANK( comm_palm, myid, ierr )
214
215    ELSE
216
217       CALL MPI_COMM_SIZE( MPI_COMM_WORLD, numprocs, ierr )
218       CALL MPI_COMM_RANK( MPI_COMM_WORLD, myid, ierr )
219!
220!--    Initialize PE topology in case of coupled atmosphere-ocean runs (comm_palm
221!--    will be splitted in init_coupling)
222       CALL init_coupling
223    ENDIF
224#endif
225
226#if defined( __openacc )
227!
228!-- Get the local MPI rank in case of CUDA aware OpenMPI. Important, if there
229!-- is more than one accelerator board on the node
230    CALL GET_ENVIRONMENT_VARIABLE('OMPI_COMM_WORLD_LOCAL_RANK',                &
231         VALUE=env_string, STATUS=env_stat )
232    READ( env_string, '(I1)' )  myid_openmpi
233    PRINT*, '### local_rank = ', myid_openmpi, '  status=',env_stat
234!
235!-- Get the number of accelerator boards per node and assign the MPI processes
236!-- to these boards
237    PRINT*, '*** ACC_DEVICE_NVIDIA = ', ACC_DEVICE_NVIDIA
238    num_acc_per_node  = ACC_GET_NUM_DEVICES( ACC_DEVICE_NVIDIA )
239    IF ( numprocs == 1  .AND.  num_acc_per_node > 0 )  num_acc_per_node = 1
240    PRINT*, '*** myid = ', myid_openmpi, ' num_acc_per_node = ', num_acc_per_node
241    acc_rank = MOD( myid_openmpi, num_acc_per_node )
242    CALL ACC_SET_DEVICE_NUM ( acc_rank, ACC_DEVICE_NVIDIA )
243!
244!-- Test output (to be removed later)
245    WRITE (*,'(A,I6,A,I3,A,I3,A,I3)') '*** Connect MPI-Task ', myid_openmpi,   &
246                                      ' to CPU ', acc_rank, ' Devices: ',      &
247                                      num_acc_per_node, ' connected to:',      &
248                                      ACC_GET_DEVICE_NUM( ACC_DEVICE_NVIDIA )
249#endif
250
251!
252!-- Ensure that OpenACC first attaches the GPU devices by copying a dummy data
253!-- region
254    !$acc data copyin( acc_dum )
255
256!
257!-- Initialize measuring of the CPU-time remaining to the run
258    CALL local_tremain_ini
259
260!
261!-- Start of total CPU time measuring.
262    CALL cpu_log( log_point(1), 'total', 'start' )
263    CALL cpu_log( log_point(2), 'initialisation', 'start' )
264
265!
266!-- Open a file for debug output
267    WRITE (myid_char,'(''_'',I6.6)')  myid
268    OPEN( 9, FILE='DEBUG'//TRIM( coupling_char )//myid_char, FORM='FORMATTED' )
269
270!
271!-- Initialize dvrp logging. Also, one PE maybe split from the global
272!-- communicator for doing the dvrp output. In that case, the number of
273!-- PEs available for PALM is reduced by one and communicator comm_palm
274!-- is changed respectively.
275#if defined( __parallel )
276    CALL MPI_COMM_RANK( comm_palm, myid, ierr )
277!
278!-- TEST OUTPUT (TO BE REMOVED)
279    WRITE(9,*) '*** coupling_mode = "', TRIM( coupling_mode ), '"'
280    CALL LOCAL_FLUSH( 9 )
281    IF ( TRIM( coupling_mode ) /= 'uncoupled' )  THEN
282       PRINT*, '*** PE', myid, ' Global target PE:', target_id, &
283               TRIM( coupling_mode )
284    ENDIF
285#endif
286
287    CALL init_dvrp_logging
288
289!
290!-- Read control parameters from NAMELIST files and read environment-variables
291    CALL parin
292
293!
294!-- Check for the user's interface version
295    IF ( user_interface_current_revision /= user_interface_required_revision )  &
296    THEN
297       message_string = 'current user-interface revision "' //                  &
298                        TRIM( user_interface_current_revision ) // '" does ' // &
299                        'not match the required revision ' //                   &
300                        TRIM( user_interface_required_revision )
301        CALL message( 'palm', 'PA0169', 1, 2, 0, 6, 0 )
302    ENDIF
303
304!
305!-- Determine processor topology and local array indices
306    CALL init_pegrid
307
308!
309!-- Generate grid parameters
310    CALL init_grid
311
312!
313!-- Initialize nudging if required
314    IF ( nudging )  THEN
315       CALL init_nudge
316    ENDIF
317
318!
319!-- Initialize reading of large scale forcing from external file - if required
320    IF ( large_scale_forcing )  THEN
321       CALL init_ls_forcing
322    ENDIF
323
324!
325!-- Check control parameters and deduce further quantities
326    CALL check_parameters
327
328!
329!-- Initialize all necessary variables
330    CALL init_3d_model
331
332!
333!-- Coupling protocol setup for nested-domain runs
334    IF ( nested_run )  THEN
335       CALL pmci_modelconfiguration
336!
337!--    Receive and interpolate initial data on client.
338!--    Client initialization must be made first if the model is both client and
339!--    server
340       CALL pmci_client_initialize
341!
342!--    Send initial condition data from server to client
343       CALL pmci_server_initialize
344!
345!--    Exchange_horiz is needed after the nest initialization
346       IF ( nest_domain )  THEN
347          CALL exchange_horiz( u, nbgp )
348          CALL exchange_horiz( v, nbgp )
349          CALL exchange_horiz( w, nbgp )
350          CALL exchange_horiz( pt, nbgp )
351          IF ( .NOT. constant_diffusion )  CALL exchange_horiz( e, nbgp )
352          IF (humidity  .OR.  passive_scalar)  THEN
353             CALL exchange_horiz( q, nbgp )
354          ENDIF
355       ENDIF
356
357    ENDIF
358
359
360!
361!-- Output of program header
362    IF ( myid == 0 )  CALL header
363
364    CALL cpu_log( log_point(2), 'initialisation', 'stop' )
365
366!
367!-- Set start time in format hh:mm:ss
368    simulated_time_chr = time_to_string( simulated_time )
369
370!
371!-- If required, output of initial arrays
372    IF ( do2d_at_begin )  THEN
373       CALL data_output_2d( 'xy', 0 )
374       CALL data_output_2d( 'xz', 0 )
375       CALL data_output_2d( 'yz', 0 )
376    ENDIF
377    IF ( do3d_at_begin )  THEN
378       CALL data_output_3d( 0 )
379    ENDIF
380
381!
382!-- Declare and initialize variables in the accelerator memory with their
383!-- host values
384    !$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 )          &
385    !$acc       copyin( tri, tric, dzu, ddzu, ddzw, dd2zu, l_grid, l_wall, ptdf_x, ptdf_y, pt_init, rdf, rdf_sc, ref_state, ug, u_init, vg, v_init, zu, zw )   &
386    !$acc       copyin( hom, ol, pt1, qs, qsws, qswst, qv1, rif_wall, shf, ts, tswst, us, usws, uswst, uv_total, vsws, vswst, z0, z0h )      &
387    !$acc       copyin( fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, nzb_diff_s_inner, nzb_diff_s_outer, nzb_diff_u )       &
388    !$acc       copyin( nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner )    &
389    !$acc       copyin( nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner )   &
390    !$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 )  &
391    !$acc       copyin( ngp_2dh, ngp_2dh_s_inner )  &
392    !$acc       copyin( weight_pres, weight_substep )
393!
394!-- Integration of the model equations using timestep-scheme
395    CALL time_integration
396
397!
398!-- If required, write binary data for restart runs
399    IF ( write_binary(1:4) == 'true' )  THEN
400
401       CALL cpu_log( log_point(22), 'write_3d_binary', 'start' )
402
403       CALL location_message( 'writing restart data', .FALSE. )
404
405       CALL check_open( 14 )
406
407       DO  i = 0, io_blocks-1
408          IF ( i == io_group )  THEN
409!
410!--          Write flow field data
411             CALL write_3d_binary
412          ENDIF
413#if defined( __parallel )
414          CALL MPI_BARRIER( comm2d, ierr )
415#endif
416       ENDDO
417
418       CALL location_message( 'finished', .TRUE. )
419
420       CALL cpu_log( log_point(22), 'write_3d_binary', 'stop' )
421
422!
423!--    If required, write particle data
424       IF ( particle_advection )  CALL lpm_write_restart_file
425    ENDIF
426
427!
428!-- If required, repeat output of header including the required CPU-time
429    IF ( myid == 0 )  CALL header
430!
431!-- If required, final user-defined actions, and
432!-- last actions on the open files and close files. Unit 14 was opened
433!-- in write_3d_binary but it is closed here, to allow writing on this
434!-- unit in routine user_last_actions.
435    CALL cpu_log( log_point(4), 'last actions', 'start' )
436    DO  i = 0, io_blocks-1
437       IF ( i == io_group )  THEN
438          CALL user_last_actions
439          IF ( write_binary(1:4) == 'true' )  CALL close_file( 14 )
440       ENDIF
441#if defined( __parallel )
442       CALL MPI_BARRIER( comm2d, ierr )
443#endif
444    ENDDO
445    CALL close_file( 0 )
446    CALL close_dvrp
447    CALL cpu_log( log_point(4), 'last actions', 'stop' )
448
449#if defined( __mpi2 )
450!
451!-- Test exchange via intercommunicator in case of a MPI-2 coupling
452    IF ( coupling_mode == 'atmosphere_to_ocean' )  THEN
453       i = 12345 + myid
454       CALL MPI_SEND( i, 1, MPI_INTEGER, myid, 11, comm_inter, ierr )
455    ELSEIF ( coupling_mode == 'ocean_to_atmosphere' )  THEN
456       CALL MPI_RECV( i, 1, MPI_INTEGER, myid, 11, comm_inter, status, ierr )
457       PRINT*, '### myid: ', myid, '   received from atmosphere:  i = ', i
458    ENDIF
459#endif
460
461!
462!-- Close the OpenACC dummy data region
463    !$acc end data
464    !$acc end data
465
466!
467!-- Take final CPU-time for CPU-time analysis
468    CALL cpu_log( log_point(1), 'total', 'stop' )
469    CALL cpu_statistics
470
471#if defined( __parallel )
472    CALL MPI_FINALIZE( ierr )
473#endif
474
475 END PROGRAM palm
Note: See TracBrowser for help on using the repository browser.