source: palm/trunk/SOURCE/palm.f90 @ 1764

Last change on this file since 1764 was 1764, checked in by raasch, 9 years ago

update of the nested domain system + some bugfixes

  • Property svn:keywords set to Id
File size: 15.0 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! -----------------
[1764]21! cpp-statements for nesting removed, communicator settings cleaned up
[1748]22!
[1321]23! Former revisions:
24! -----------------
25! $Id: palm.f90 1764 2016-02-28 12:45:19Z raasch $
26!
[1763]27! 1762 2016-02-25 12:31:13Z hellstea
28! Introduction of nested domain feature
29!
[1748]30! 1747 2016-02-08 12:25:53Z raasch
31! OpenACC-adjustment for new surface layer parameterization
32!
[1683]33! 1682 2015-10-07 23:56:08Z knoop
34! Code annotations made doxygen readable
35!
[1669]36! 1668 2015-09-23 13:45:36Z raasch
37! warning replaced by abort in case of failed user interface check
38!
[1667]39! 1666 2015-09-23 07:31:10Z raasch
40! check for user's interface version added
41!
[1483]42! 1482 2014-10-18 12:34:45Z raasch
43! adjustments for using CUDA-aware OpenMPI
44!
[1469]45! 1468 2014-09-24 14:06:57Z maronga
46! Adapted for use on up to 6-digit processor cores
47!
[1403]48! 1402 2014-05-09 14:25:13Z raasch
49! location messages added
50!
[1375]51! 1374 2014-04-25 12:55:07Z raasch
52! bugfix: various modules added
53!
[1321]54! 1320 2014-03-20 08:40:49Z raasch
[1320]55! ONLY-attribute added to USE-statements,
56! kind-parameters added to all INTEGER and REAL declaration statements,
57! kinds are defined in new module kinds,
58! old module precision_kind is removed,
59! revision history before 2012 removed,
60! comment fields (!:) to be used for variable explanations added to
61! all variable declaration statements
[77]62!
[1319]63! 1318 2014-03-17 13:35:16Z raasch
64! module interfaces removed
65!
[1242]66! 1241 2013-10-30 11:36:58Z heinze
67! initialization of nuding and large scale forcing from external file
68!
[1222]69! 1221 2013-09-10 08:59:13Z raasch
70! +wall_flags_00, rflags_invers, rflags_s_inner in copyin statement
71!
[1213]72! 1212 2013-08-15 08:46:27Z raasch
73! +tri in copyin statement
74!
[1182]75! 1179 2013-06-14 05:57:58Z raasch
76! ref_state added to copyin-list
77!
[1114]78! 1113 2013-03-10 02:48:14Z raasch
79! openACC statements modified
80!
[1112]81! 1111 2013-03-08 23:54:10Z raasch
82! openACC statements updated
83!
[1093]84! 1092 2013-02-02 11:24:22Z raasch
85! unused variables removed
86!
[1037]87! 1036 2012-10-22 13:43:42Z raasch
88! code put under GPL (PALM 3.9)
89!
[1017]90! 1015 2012-09-27 09:23:24Z raasch
[863]91! Version number changed from 3.8 to 3.8a.
[1017]92! OpenACC statements added + code changes required for GPU optimization
[863]93!
[850]94! 849 2012-03-15 10:35:09Z raasch
95! write_particles renamed lpm_write_restart_file
96!
[1]97! Revision 1.1  1997/07/24 11:23:35  raasch
98! Initial revision
99!
100!
101! Description:
102! ------------
[1682]103!> Large-Eddy Simulation (LES) model for the convective boundary layer,
104!> optimized for use on parallel machines (implementation realized using the
105!> Message Passing Interface (MPI)). The model can also be run on vector machines
106!> (less well optimized) and workstations. Versions for the different types of
107!> machines are controlled via cpp-directives.
108!> Model runs are only feasible using the ksh-script mrun.
[1]109!------------------------------------------------------------------------------!
[1682]110 PROGRAM palm
111 
[1]112
[1374]113    USE arrays_3d
[1]114
[1320]115    USE control_parameters,                                                    &
116        ONLY:  coupling_char, coupling_mode, do2d_at_begin, do3d_at_begin,     &
[1666]117               io_blocks, io_group, large_scale_forcing, message_string,       &
[1762]118               nest_domain, nudging, simulated_time, simulated_time_chr,       &
[1666]119               user_interface_current_revision,                                &
120               user_interface_required_revision, version, wall_heatflux,       &
121               write_binary
[1320]122
123    USE cpulog,                                                                &
124        ONLY:  cpu_log, log_point, cpu_statistics
125
[1374]126    USE grid_variables,                                                        &
127        ONLY:  fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, wall_e_x, wall_e_y, &
128               wall_u, wall_v, wall_w_x, wall_w_y
129
130    USE indices,                                                               &
[1666]131        ONLY:  ngp_2dh, ngp_2dh_s_inner, nzb_diff_s_inner, nzb_diff_s_outer,   &
132               nzb_diff_u, nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner,  &
133               nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner,             &
134               nzb_w_outer, rflags_invers, rflags_s_inner, wall_flags_0,       &
135               wall_flags_00
[1374]136
[1320]137    USE kinds
138
139    USE ls_forcing_mod,                                                        &
140        ONLY:  init_ls_forcing
141
142    USE nudge_mod,                                                             &
143        ONLY:  init_nudge
144
145    USE particle_attributes,                                                   &
146        ONLY:  particle_advection
147
[1]148    USE pegrid
149
[1762]150    USE pmc_interface,                                                         &
[1764]151        ONLY:  cpl_id, nested_run, pmci_init, pmci_modelconfiguration
[1762]152
[1374]153    USE statistics,                                                            &
154        ONLY:  hom, rmask, weight_pres, weight_substep
155
[1747]156    USE surface_layer_fluxes_mod,                                              &
157        ONLY:  pt1, qv1, uv_total
158
[1015]159#if defined( __openacc )
160    USE OPENACC
161#endif
162
[1]163    IMPLICIT NONE
164
165!
166!-- Local variables
[1682]167    CHARACTER(LEN=9)  ::  time_to_string  !<
168    CHARACTER(LEN=10) ::  env_string      !< to store string of environment var
169    INTEGER(iwp)      ::  env_stat        !< to hold status of GET_ENV
170    INTEGER(iwp)      ::  i               !<
171    INTEGER(iwp)      ::  myid_openmpi    !< OpenMPI local rank for CUDA aware MPI
[1015]172#if defined( __openacc )
[1682]173    REAL(wp), DIMENSION(100) ::  acc_dum     !<
[1015]174#endif
[1]175
[1539]176    version = 'PALM 4.0'
[1666]177    user_interface_required_revision = 'r1663'
[75]178
[1]179#if defined( __parallel )
180!
181!-- MPI initialisation. comm2d is preliminary set, because
182!-- it will be defined in init_pegrid but is used before in cpu_log.
183    CALL MPI_INIT( ierr )
[1762]184
185!
186!-- Initialize the coupling for nested-domain runs
[1764]187!-- comm_palm is the communicator which includes all PEs (MPI processes)
188!-- available for this (nested) model. If it is not a nested run, comm_palm
189!-- is returned as MPI_COMM_WORLD
[1762]190    CALL pmci_init( comm_palm )
191    comm2d = comm_palm
[1764]192!
193!-- Get the (preliminary) number of MPI processes and the local PE-id (in case
194!-- of a further communicator splitting in init_coupling, these numbers will
195!-- be changed in init_pegrid).
196    IF ( nested_run )  THEN
197!--    TO_DO: move the following two settings somewehere to the pmc_interface
198       IF ( cpl_id >= 2 )  THEN
199          nest_domain = .TRUE.
200          WRITE( coupling_char, '(A1,I1.1)') '_', cpl_id
201       ENDIF
[1762]202
[1764]203       CALL MPI_COMM_SIZE( comm_palm, numprocs, ierr )
204       CALL MPI_COMM_RANK( comm_palm, myid, ierr )
[1762]205
[1764]206    ELSE
207
208       CALL MPI_COMM_SIZE( MPI_COMM_WORLD, numprocs, ierr )
209       CALL MPI_COMM_RANK( MPI_COMM_WORLD, myid, ierr )
[1]210!
[1764]211!--    Initialize PE topology in case of coupled atmosphere-ocean runs (comm_palm
212!--    will be splitted in init_coupling)
213       CALL init_coupling
214    ENDIF
[102]215#endif
216
[1015]217#if defined( __openacc )
[102]218!
[1482]219!-- Get the local MPI rank in case of CUDA aware OpenMPI. Important, if there
220!-- is more than one accelerator board on the node
221    CALL GET_ENVIRONMENT_VARIABLE('OMPI_COMM_WORLD_LOCAL_RANK',                &
222         VALUE=env_string, STATUS=env_stat )
223    READ( env_string, '(I1)' )  myid_openmpi
224    PRINT*, '### local_rank = ', myid_openmpi, '  status=',env_stat
225!
[1015]226!-- Get the number of accelerator boards per node and assign the MPI processes
227!-- to these boards
[1092]228    PRINT*, '*** ACC_DEVICE_NVIDIA = ', ACC_DEVICE_NVIDIA
[1015]229    num_acc_per_node  = ACC_GET_NUM_DEVICES( ACC_DEVICE_NVIDIA )
[1092]230    IF ( numprocs == 1  .AND.  num_acc_per_node > 0 )  num_acc_per_node = 1
[1482]231    PRINT*, '*** myid = ', myid_openmpi, ' num_acc_per_node = ', num_acc_per_node
232    acc_rank = MOD( myid_openmpi, num_acc_per_node )
[1015]233    CALL ACC_SET_DEVICE_NUM ( acc_rank, ACC_DEVICE_NVIDIA )
234!
235!-- Test output (to be removed later)
[1482]236    WRITE (*,'(A,I6,A,I3,A,I3,A,I3)') '*** Connect MPI-Task ', myid_openmpi,   &
237                                      ' to CPU ', acc_rank, ' Devices: ',      &
238                                      num_acc_per_node, ' connected to:',      &
[1015]239                                      ACC_GET_DEVICE_NUM( ACC_DEVICE_NVIDIA )
240#endif
[1092]241
[1015]242!
243!-- Ensure that OpenACC first attaches the GPU devices by copying a dummy data
244!-- region
245    !$acc data copyin( acc_dum )
246
247!
[1]248!-- Initialize measuring of the CPU-time remaining to the run
249    CALL local_tremain_ini
250
251!
252!-- Start of total CPU time measuring.
253    CALL cpu_log( log_point(1), 'total', 'start' )
254    CALL cpu_log( log_point(2), 'initialisation', 'start' )
255
256!
[206]257!-- Open a file for debug output
[1468]258    WRITE (myid_char,'(''_'',I6.6)')  myid
[206]259    OPEN( 9, FILE='DEBUG'//TRIM( coupling_char )//myid_char, FORM='FORMATTED' )
260
261!
[1]262!-- Initialize dvrp logging. Also, one PE maybe split from the global
263!-- communicator for doing the dvrp output. In that case, the number of
264!-- PEs available for PALM is reduced by one and communicator comm_palm
265!-- is changed respectively.
266#if defined( __parallel )
267    CALL MPI_COMM_RANK( comm_palm, myid, ierr )
268!
[102]269!-- TEST OUTPUT (TO BE REMOVED)
270    WRITE(9,*) '*** coupling_mode = "', TRIM( coupling_mode ), '"'
271    CALL LOCAL_FLUSH( 9 )
[215]272    IF ( TRIM( coupling_mode ) /= 'uncoupled' )  THEN
273       PRINT*, '*** PE', myid, ' Global target PE:', target_id, &
274               TRIM( coupling_mode )
275    ENDIF
[102]276#endif
277
[108]278    CALL init_dvrp_logging
279
[102]280!
[108]281!-- Read control parameters from NAMELIST files and read environment-variables
282    CALL parin
283
284!
[1666]285!-- Check for the user's interface version
286    IF ( user_interface_current_revision /= user_interface_required_revision )  &
287    THEN
288       message_string = 'current user-interface revision "' //                  &
289                        TRIM( user_interface_current_revision ) // '" does ' // &
290                        'not match the required revision ' //                   &
291                        TRIM( user_interface_required_revision )
[1668]292        CALL message( 'palm', 'PA0169', 1, 2, 0, 6, 0 )
[1666]293    ENDIF
294
295!
[108]296!-- Determine processor topology and local array indices
297    CALL init_pegrid
298
299!
[1]300!-- Generate grid parameters
301    CALL init_grid
302
303!
[1241]304!-- Initialize nudging if required
305    IF ( nudging )  THEN
306       CALL init_nudge
307    ENDIF
308
309!
310!-- Initialize reading of large scale forcing from external file - if required
311    IF ( large_scale_forcing )  THEN
312       CALL init_ls_forcing
313    ENDIF
314
315!
[1]316!-- Check control parameters and deduce further quantities
317    CALL check_parameters
318
319!
320!-- Initialize all necessary variables
321    CALL init_3d_model
322
323!
[1762]324!-- Coupling protocol setup for nested-domain runs
[1764]325    IF ( nested_run )  THEN
326       CALL pmci_modelconfiguration
327    ENDIF
[1762]328
329!
[1]330!-- Output of program header
331    IF ( myid == 0 )  CALL header
332
333    CALL cpu_log( log_point(2), 'initialisation', 'stop' )
334
335!
336!-- Set start time in format hh:mm:ss
337    simulated_time_chr = time_to_string( simulated_time )
338
339!
340!-- If required, output of initial arrays
341    IF ( do2d_at_begin )  THEN
342       CALL data_output_2d( 'xy', 0 )
343       CALL data_output_2d( 'xz', 0 )
344       CALL data_output_2d( 'yz', 0 )
345    ENDIF
346    IF ( do3d_at_begin )  THEN
347       CALL data_output_3d( 0 )
348    ENDIF
349
350!
[1015]351!-- Declare and initialize variables in the accelerator memory with their
352!-- host values
[1113]353    !$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]354    !$acc       copyin( tri, tric, dzu, ddzu, ddzw, dd2zu, l_grid, l_wall, ptdf_x, ptdf_y, pt_init, rdf, rdf_sc, ref_state, ug, u_init, vg, v_init, zu, zw )   &
[1747]355    !$acc       copyin( hom, ol, pt1, qs, qsws, qswst, qv1, rif_wall, shf, ts, tswst, us, usws, uswst, uv_total, vsws, vswst, z0, z0h )      &
[1015]356    !$acc       copyin( fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, nzb_diff_s_inner, nzb_diff_s_outer, nzb_diff_u )       &
357    !$acc       copyin( nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner )    &
358    !$acc       copyin( nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner )   &
[1221]359    !$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 )  &
360    !$acc       copyin( ngp_2dh, ngp_2dh_s_inner )  &
[1113]361    !$acc       copyin( weight_pres, weight_substep )
[1015]362!
[495]363!-- Integration of the model equations using timestep-scheme
[1]364    CALL time_integration
365
366!
[495]367!-- If required, write binary data for restart runs
368    IF ( write_binary(1:4) == 'true' )  THEN
[759]369
370       CALL cpu_log( log_point(22), 'write_3d_binary', 'start' )
371
[1402]372       CALL location_message( 'writing restart data', .FALSE. )
373
[759]374       CALL check_open( 14 )
375
376       DO  i = 0, io_blocks-1
377          IF ( i == io_group )  THEN
[1]378!
[759]379!--          Write flow field data
380             CALL write_3d_binary
381          ENDIF
382#if defined( __parallel )
383          CALL MPI_BARRIER( comm2d, ierr )
384#endif
385       ENDDO
386
[1402]387       CALL location_message( 'finished', .TRUE. )
388
[759]389       CALL cpu_log( log_point(22), 'write_3d_binary', 'stop' )
390
[495]391!
392!--    If required, write particle data
[849]393       IF ( particle_advection )  CALL lpm_write_restart_file
[495]394    ENDIF
[1]395
396!
397!-- If required, repeat output of header including the required CPU-time
398    IF ( myid == 0 )  CALL header
399!
400!-- If required, final user-defined actions, and
401!-- last actions on the open files and close files. Unit 14 was opened
402!-- in write_3d_binary but it is closed here, to allow writing on this
403!-- unit in routine user_last_actions.
404    CALL cpu_log( log_point(4), 'last actions', 'start' )
[759]405    DO  i = 0, io_blocks-1
406       IF ( i == io_group )  THEN
407          CALL user_last_actions
408          IF ( write_binary(1:4) == 'true' )  CALL close_file( 14 )
409       ENDIF
410#if defined( __parallel )
411       CALL MPI_BARRIER( comm2d, ierr )
412#endif
413    ENDDO
[1]414    CALL close_file( 0 )
415    CALL close_dvrp
416    CALL cpu_log( log_point(4), 'last actions', 'stop' )
417
[102]418#if defined( __mpi2 )
[1]419!
[206]420!-- Test exchange via intercommunicator in case of a MPI-2 coupling
[102]421    IF ( coupling_mode == 'atmosphere_to_ocean' )  THEN
422       i = 12345 + myid
423       CALL MPI_SEND( i, 1, MPI_INTEGER, myid, 11, comm_inter, ierr )
424    ELSEIF ( coupling_mode == 'ocean_to_atmosphere' )  THEN
425       CALL MPI_RECV( i, 1, MPI_INTEGER, myid, 11, comm_inter, status, ierr )
426       PRINT*, '### myid: ', myid, '   received from atmosphere:  i = ', i
427    ENDIF
428#endif
429
430!
[1015]431!-- Close the OpenACC dummy data region
432    !$acc end data
433    !$acc end data
434
435!
[1]436!-- Take final CPU-time for CPU-time analysis
437    CALL cpu_log( log_point(1), 'total', 'stop' )
438    CALL cpu_statistics
439
440#if defined( __parallel )
441    CALL MPI_FINALIZE( ierr )
442#endif
443
444 END PROGRAM palm
Note: See TracBrowser for help on using the repository browser.