source: palm/trunk/SOURCE/palm.f90 @ 1780

Last change on this file since 1780 was 1780, checked in by raasch, 8 years ago

last commit documented

  • 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! -----------------
[1748]21!
[1780]22!
[1321]23! Former revisions:
24! -----------------
25! $Id: palm.f90 1780 2016-03-03 08:39:40Z raasch $
26!
[1780]27! 1779 2016-03-03 08:01:28Z raasch
28! setting of nest_domain and coupling_char moved to the pmci
29!
[1765]30! 1764 2016-02-28 12:45:19Z raasch
31! cpp-statements for nesting removed, communicator settings cleaned up
32!
[1763]33! 1762 2016-02-25 12:31:13Z hellstea
34! Introduction of nested domain feature
35!
[1748]36! 1747 2016-02-08 12:25:53Z raasch
37! OpenACC-adjustment for new surface layer parameterization
38!
[1683]39! 1682 2015-10-07 23:56:08Z knoop
40! Code annotations made doxygen readable
41!
[1669]42! 1668 2015-09-23 13:45:36Z raasch
43! warning replaced by abort in case of failed user interface check
44!
[1667]45! 1666 2015-09-23 07:31:10Z raasch
46! check for user's interface version added
47!
[1483]48! 1482 2014-10-18 12:34:45Z raasch
49! adjustments for using CUDA-aware OpenMPI
50!
[1469]51! 1468 2014-09-24 14:06:57Z maronga
52! Adapted for use on up to 6-digit processor cores
53!
[1403]54! 1402 2014-05-09 14:25:13Z raasch
55! location messages added
56!
[1375]57! 1374 2014-04-25 12:55:07Z raasch
58! bugfix: various modules added
59!
[1321]60! 1320 2014-03-20 08:40:49Z raasch
[1320]61! ONLY-attribute added to USE-statements,
62! kind-parameters added to all INTEGER and REAL declaration statements,
63! kinds are defined in new module kinds,
64! old module precision_kind is removed,
65! revision history before 2012 removed,
66! comment fields (!:) to be used for variable explanations added to
67! all variable declaration statements
[77]68!
[1319]69! 1318 2014-03-17 13:35:16Z raasch
70! module interfaces removed
71!
[1242]72! 1241 2013-10-30 11:36:58Z heinze
73! initialization of nuding and large scale forcing from external file
74!
[1222]75! 1221 2013-09-10 08:59:13Z raasch
76! +wall_flags_00, rflags_invers, rflags_s_inner in copyin statement
77!
[1213]78! 1212 2013-08-15 08:46:27Z raasch
79! +tri in copyin statement
80!
[1182]81! 1179 2013-06-14 05:57:58Z raasch
82! ref_state added to copyin-list
83!
[1114]84! 1113 2013-03-10 02:48:14Z raasch
85! openACC statements modified
86!
[1112]87! 1111 2013-03-08 23:54:10Z raasch
88! openACC statements updated
89!
[1093]90! 1092 2013-02-02 11:24:22Z raasch
91! unused variables removed
92!
[1037]93! 1036 2012-10-22 13:43:42Z raasch
94! code put under GPL (PALM 3.9)
95!
[1017]96! 1015 2012-09-27 09:23:24Z raasch
[863]97! Version number changed from 3.8 to 3.8a.
[1017]98! OpenACC statements added + code changes required for GPU optimization
[863]99!
[850]100! 849 2012-03-15 10:35:09Z raasch
101! write_particles renamed lpm_write_restart_file
102!
[1]103! Revision 1.1  1997/07/24 11:23:35  raasch
104! Initial revision
105!
106!
107! Description:
108! ------------
[1682]109!> Large-Eddy Simulation (LES) model for the convective boundary layer,
110!> optimized for use on parallel machines (implementation realized using the
111!> Message Passing Interface (MPI)). The model can also be run on vector machines
112!> (less well optimized) and workstations. Versions for the different types of
113!> machines are controlled via cpp-directives.
114!> Model runs are only feasible using the ksh-script mrun.
[1]115!------------------------------------------------------------------------------!
[1682]116 PROGRAM palm
117 
[1]118
[1374]119    USE arrays_3d
[1]120
[1320]121    USE control_parameters,                                                    &
122        ONLY:  coupling_char, coupling_mode, do2d_at_begin, do3d_at_begin,     &
[1666]123               io_blocks, io_group, large_scale_forcing, message_string,       &
[1762]124               nest_domain, nudging, simulated_time, simulated_time_chr,       &
[1666]125               user_interface_current_revision,                                &
126               user_interface_required_revision, version, wall_heatflux,       &
127               write_binary
[1320]128
129    USE cpulog,                                                                &
130        ONLY:  cpu_log, log_point, cpu_statistics
131
[1374]132    USE grid_variables,                                                        &
133        ONLY:  fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, wall_e_x, wall_e_y, &
134               wall_u, wall_v, wall_w_x, wall_w_y
135
136    USE indices,                                                               &
[1666]137        ONLY:  ngp_2dh, ngp_2dh_s_inner, nzb_diff_s_inner, nzb_diff_s_outer,   &
138               nzb_diff_u, nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner,  &
139               nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner,             &
140               nzb_w_outer, rflags_invers, rflags_s_inner, wall_flags_0,       &
141               wall_flags_00
[1374]142
[1320]143    USE kinds
144
145    USE ls_forcing_mod,                                                        &
146        ONLY:  init_ls_forcing
147
148    USE nudge_mod,                                                             &
149        ONLY:  init_nudge
150
151    USE particle_attributes,                                                   &
152        ONLY:  particle_advection
153
[1]154    USE pegrid
155
[1762]156    USE pmc_interface,                                                         &
[1764]157        ONLY:  cpl_id, nested_run, pmci_init, pmci_modelconfiguration
[1762]158
[1374]159    USE statistics,                                                            &
160        ONLY:  hom, rmask, weight_pres, weight_substep
161
[1747]162    USE surface_layer_fluxes_mod,                                              &
163        ONLY:  pt1, qv1, uv_total
164
[1015]165#if defined( __openacc )
166    USE OPENACC
167#endif
168
[1]169    IMPLICIT NONE
170
171!
172!-- Local variables
[1682]173    CHARACTER(LEN=9)  ::  time_to_string  !<
174    CHARACTER(LEN=10) ::  env_string      !< to store string of environment var
175    INTEGER(iwp)      ::  env_stat        !< to hold status of GET_ENV
176    INTEGER(iwp)      ::  i               !<
177    INTEGER(iwp)      ::  myid_openmpi    !< OpenMPI local rank for CUDA aware MPI
[1015]178#if defined( __openacc )
[1682]179    REAL(wp), DIMENSION(100) ::  acc_dum     !<
[1015]180#endif
[1]181
[1539]182    version = 'PALM 4.0'
[1666]183    user_interface_required_revision = 'r1663'
[75]184
[1]185#if defined( __parallel )
186!
187!-- MPI initialisation. comm2d is preliminary set, because
188!-- it will be defined in init_pegrid but is used before in cpu_log.
189    CALL MPI_INIT( ierr )
[1762]190
191!
192!-- Initialize the coupling for nested-domain runs
[1764]193!-- comm_palm is the communicator which includes all PEs (MPI processes)
194!-- available for this (nested) model. If it is not a nested run, comm_palm
195!-- is returned as MPI_COMM_WORLD
[1762]196    CALL pmci_init( comm_palm )
197    comm2d = comm_palm
[1764]198!
199!-- Get the (preliminary) number of MPI processes and the local PE-id (in case
200!-- of a further communicator splitting in init_coupling, these numbers will
201!-- be changed in init_pegrid).
202    IF ( nested_run )  THEN
[1762]203
[1764]204       CALL MPI_COMM_SIZE( comm_palm, numprocs, ierr )
205       CALL MPI_COMM_RANK( comm_palm, myid, ierr )
[1762]206
[1764]207    ELSE
208
209       CALL MPI_COMM_SIZE( MPI_COMM_WORLD, numprocs, ierr )
210       CALL MPI_COMM_RANK( MPI_COMM_WORLD, myid, ierr )
[1]211!
[1764]212!--    Initialize PE topology in case of coupled atmosphere-ocean runs (comm_palm
213!--    will be splitted in init_coupling)
214       CALL init_coupling
215    ENDIF
[102]216#endif
217
[1015]218#if defined( __openacc )
[102]219!
[1482]220!-- Get the local MPI rank in case of CUDA aware OpenMPI. Important, if there
221!-- is more than one accelerator board on the node
222    CALL GET_ENVIRONMENT_VARIABLE('OMPI_COMM_WORLD_LOCAL_RANK',                &
223         VALUE=env_string, STATUS=env_stat )
224    READ( env_string, '(I1)' )  myid_openmpi
225    PRINT*, '### local_rank = ', myid_openmpi, '  status=',env_stat
226!
[1015]227!-- Get the number of accelerator boards per node and assign the MPI processes
228!-- to these boards
[1092]229    PRINT*, '*** ACC_DEVICE_NVIDIA = ', ACC_DEVICE_NVIDIA
[1015]230    num_acc_per_node  = ACC_GET_NUM_DEVICES( ACC_DEVICE_NVIDIA )
[1092]231    IF ( numprocs == 1  .AND.  num_acc_per_node > 0 )  num_acc_per_node = 1
[1482]232    PRINT*, '*** myid = ', myid_openmpi, ' num_acc_per_node = ', num_acc_per_node
233    acc_rank = MOD( myid_openmpi, num_acc_per_node )
[1015]234    CALL ACC_SET_DEVICE_NUM ( acc_rank, ACC_DEVICE_NVIDIA )
235!
236!-- Test output (to be removed later)
[1482]237    WRITE (*,'(A,I6,A,I3,A,I3,A,I3)') '*** Connect MPI-Task ', myid_openmpi,   &
238                                      ' to CPU ', acc_rank, ' Devices: ',      &
239                                      num_acc_per_node, ' connected to:',      &
[1015]240                                      ACC_GET_DEVICE_NUM( ACC_DEVICE_NVIDIA )
241#endif
[1092]242
[1015]243!
244!-- Ensure that OpenACC first attaches the GPU devices by copying a dummy data
245!-- region
246    !$acc data copyin( acc_dum )
247
248!
[1]249!-- Initialize measuring of the CPU-time remaining to the run
250    CALL local_tremain_ini
251
252!
253!-- Start of total CPU time measuring.
254    CALL cpu_log( log_point(1), 'total', 'start' )
255    CALL cpu_log( log_point(2), 'initialisation', 'start' )
256
257!
[206]258!-- Open a file for debug output
[1468]259    WRITE (myid_char,'(''_'',I6.6)')  myid
[206]260    OPEN( 9, FILE='DEBUG'//TRIM( coupling_char )//myid_char, FORM='FORMATTED' )
261
262!
[1]263!-- Initialize dvrp logging. Also, one PE maybe split from the global
264!-- communicator for doing the dvrp output. In that case, the number of
265!-- PEs available for PALM is reduced by one and communicator comm_palm
266!-- is changed respectively.
267#if defined( __parallel )
268    CALL MPI_COMM_RANK( comm_palm, myid, ierr )
269!
[102]270!-- TEST OUTPUT (TO BE REMOVED)
271    WRITE(9,*) '*** coupling_mode = "', TRIM( coupling_mode ), '"'
272    CALL LOCAL_FLUSH( 9 )
[215]273    IF ( TRIM( coupling_mode ) /= 'uncoupled' )  THEN
274       PRINT*, '*** PE', myid, ' Global target PE:', target_id, &
275               TRIM( coupling_mode )
276    ENDIF
[102]277#endif
278
[108]279    CALL init_dvrp_logging
280
[102]281!
[108]282!-- Read control parameters from NAMELIST files and read environment-variables
283    CALL parin
284
285!
[1666]286!-- Check for the user's interface version
287    IF ( user_interface_current_revision /= user_interface_required_revision )  &
288    THEN
289       message_string = 'current user-interface revision "' //                  &
290                        TRIM( user_interface_current_revision ) // '" does ' // &
291                        'not match the required revision ' //                   &
292                        TRIM( user_interface_required_revision )
[1668]293        CALL message( 'palm', 'PA0169', 1, 2, 0, 6, 0 )
[1666]294    ENDIF
295
296!
[108]297!-- Determine processor topology and local array indices
298    CALL init_pegrid
299
300!
[1]301!-- Generate grid parameters
302    CALL init_grid
303
304!
[1241]305!-- Initialize nudging if required
306    IF ( nudging )  THEN
307       CALL init_nudge
308    ENDIF
309
310!
311!-- Initialize reading of large scale forcing from external file - if required
312    IF ( large_scale_forcing )  THEN
313       CALL init_ls_forcing
314    ENDIF
315
316!
[1]317!-- Check control parameters and deduce further quantities
318    CALL check_parameters
319
320!
321!-- Initialize all necessary variables
322    CALL init_3d_model
323
324!
[1762]325!-- Coupling protocol setup for nested-domain runs
[1764]326    IF ( nested_run )  THEN
327       CALL pmci_modelconfiguration
328    ENDIF
[1762]329
330!
[1]331!-- Output of program header
332    IF ( myid == 0 )  CALL header
333
334    CALL cpu_log( log_point(2), 'initialisation', 'stop' )
335
336!
337!-- Set start time in format hh:mm:ss
338    simulated_time_chr = time_to_string( simulated_time )
339
340!
341!-- If required, output of initial arrays
342    IF ( do2d_at_begin )  THEN
343       CALL data_output_2d( 'xy', 0 )
344       CALL data_output_2d( 'xz', 0 )
345       CALL data_output_2d( 'yz', 0 )
346    ENDIF
347    IF ( do3d_at_begin )  THEN
348       CALL data_output_3d( 0 )
349    ENDIF
350
351!
[1015]352!-- Declare and initialize variables in the accelerator memory with their
353!-- host values
[1113]354    !$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]355    !$acc       copyin( tri, tric, dzu, ddzu, ddzw, dd2zu, l_grid, l_wall, ptdf_x, ptdf_y, pt_init, rdf, rdf_sc, ref_state, ug, u_init, vg, v_init, zu, zw )   &
[1747]356    !$acc       copyin( hom, ol, pt1, qs, qsws, qswst, qv1, rif_wall, shf, ts, tswst, us, usws, uswst, uv_total, vsws, vswst, z0, z0h )      &
[1015]357    !$acc       copyin( fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, nzb_diff_s_inner, nzb_diff_s_outer, nzb_diff_u )       &
358    !$acc       copyin( nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner )    &
359    !$acc       copyin( nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner )   &
[1221]360    !$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 )  &
361    !$acc       copyin( ngp_2dh, ngp_2dh_s_inner )  &
[1113]362    !$acc       copyin( weight_pres, weight_substep )
[1015]363!
[495]364!-- Integration of the model equations using timestep-scheme
[1]365    CALL time_integration
366
367!
[495]368!-- If required, write binary data for restart runs
369    IF ( write_binary(1:4) == 'true' )  THEN
[759]370
371       CALL cpu_log( log_point(22), 'write_3d_binary', 'start' )
372
[1402]373       CALL location_message( 'writing restart data', .FALSE. )
374
[759]375       CALL check_open( 14 )
376
377       DO  i = 0, io_blocks-1
378          IF ( i == io_group )  THEN
[1]379!
[759]380!--          Write flow field data
381             CALL write_3d_binary
382          ENDIF
383#if defined( __parallel )
384          CALL MPI_BARRIER( comm2d, ierr )
385#endif
386       ENDDO
387
[1402]388       CALL location_message( 'finished', .TRUE. )
389
[759]390       CALL cpu_log( log_point(22), 'write_3d_binary', 'stop' )
391
[495]392!
393!--    If required, write particle data
[849]394       IF ( particle_advection )  CALL lpm_write_restart_file
[495]395    ENDIF
[1]396
397!
398!-- If required, repeat output of header including the required CPU-time
399    IF ( myid == 0 )  CALL header
400!
401!-- If required, final user-defined actions, and
402!-- last actions on the open files and close files. Unit 14 was opened
403!-- in write_3d_binary but it is closed here, to allow writing on this
404!-- unit in routine user_last_actions.
405    CALL cpu_log( log_point(4), 'last actions', 'start' )
[759]406    DO  i = 0, io_blocks-1
407       IF ( i == io_group )  THEN
408          CALL user_last_actions
409          IF ( write_binary(1:4) == 'true' )  CALL close_file( 14 )
410       ENDIF
411#if defined( __parallel )
412       CALL MPI_BARRIER( comm2d, ierr )
413#endif
414    ENDDO
[1]415    CALL close_file( 0 )
416    CALL close_dvrp
417    CALL cpu_log( log_point(4), 'last actions', 'stop' )
418
[102]419#if defined( __mpi2 )
[1]420!
[206]421!-- Test exchange via intercommunicator in case of a MPI-2 coupling
[102]422    IF ( coupling_mode == 'atmosphere_to_ocean' )  THEN
423       i = 12345 + myid
424       CALL MPI_SEND( i, 1, MPI_INTEGER, myid, 11, comm_inter, ierr )
425    ELSEIF ( coupling_mode == 'ocean_to_atmosphere' )  THEN
426       CALL MPI_RECV( i, 1, MPI_INTEGER, myid, 11, comm_inter, status, ierr )
427       PRINT*, '### myid: ', myid, '   received from atmosphere:  i = ', i
428    ENDIF
429#endif
430
431!
[1015]432!-- Close the OpenACC dummy data region
433    !$acc end data
434    !$acc end data
435
436!
[1]437!-- Take final CPU-time for CPU-time analysis
438    CALL cpu_log( log_point(1), 'total', 'stop' )
439    CALL cpu_statistics
440
441#if defined( __parallel )
442    CALL MPI_FINALIZE( ierr )
443#endif
444
445 END PROGRAM palm
Note: See TracBrowser for help on using the repository browser.