source: palm/trunk/SOURCE/palm.f90 @ 1752

Last change on this file since 1752 was 1748, checked in by raasch, 9 years ago

last commit documented

  • Property svn:keywords set to Id
File size: 13.7 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! -----------------
[1683]21!
[1748]22!
[1321]23! Former revisions:
24! -----------------
25! $Id: palm.f90 1748 2016-02-08 12:38:17Z maronga $
26!
[1748]27! 1747 2016-02-08 12:25:53Z raasch
28! OpenACC-adjustment for new surface layer parameterization
29!
[1683]30! 1682 2015-10-07 23:56:08Z knoop
31! Code annotations made doxygen readable
32!
[1669]33! 1668 2015-09-23 13:45:36Z raasch
34! warning replaced by abort in case of failed user interface check
35!
[1667]36! 1666 2015-09-23 07:31:10Z raasch
37! check for user's interface version added
38!
[1483]39! 1482 2014-10-18 12:34:45Z raasch
40! adjustments for using CUDA-aware OpenMPI
41!
[1469]42! 1468 2014-09-24 14:06:57Z maronga
43! Adapted for use on up to 6-digit processor cores
44!
[1403]45! 1402 2014-05-09 14:25:13Z raasch
46! location messages added
47!
[1375]48! 1374 2014-04-25 12:55:07Z raasch
49! bugfix: various modules added
50!
[1321]51! 1320 2014-03-20 08:40:49Z raasch
[1320]52! ONLY-attribute added to USE-statements,
53! kind-parameters added to all INTEGER and REAL declaration statements,
54! kinds are defined in new module kinds,
55! old module precision_kind is removed,
56! revision history before 2012 removed,
57! comment fields (!:) to be used for variable explanations added to
58! all variable declaration statements
[77]59!
[1319]60! 1318 2014-03-17 13:35:16Z raasch
61! module interfaces removed
62!
[1242]63! 1241 2013-10-30 11:36:58Z heinze
64! initialization of nuding and large scale forcing from external file
65!
[1222]66! 1221 2013-09-10 08:59:13Z raasch
67! +wall_flags_00, rflags_invers, rflags_s_inner in copyin statement
68!
[1213]69! 1212 2013-08-15 08:46:27Z raasch
70! +tri in copyin statement
71!
[1182]72! 1179 2013-06-14 05:57:58Z raasch
73! ref_state added to copyin-list
74!
[1114]75! 1113 2013-03-10 02:48:14Z raasch
76! openACC statements modified
77!
[1112]78! 1111 2013-03-08 23:54:10Z raasch
79! openACC statements updated
80!
[1093]81! 1092 2013-02-02 11:24:22Z raasch
82! unused variables removed
83!
[1037]84! 1036 2012-10-22 13:43:42Z raasch
85! code put under GPL (PALM 3.9)
86!
[1017]87! 1015 2012-09-27 09:23:24Z raasch
[863]88! Version number changed from 3.8 to 3.8a.
[1017]89! OpenACC statements added + code changes required for GPU optimization
[863]90!
[850]91! 849 2012-03-15 10:35:09Z raasch
92! write_particles renamed lpm_write_restart_file
93!
[1]94! Revision 1.1  1997/07/24 11:23:35  raasch
95! Initial revision
96!
97!
98! Description:
99! ------------
[1682]100!> Large-Eddy Simulation (LES) model for the convective boundary layer,
101!> optimized for use on parallel machines (implementation realized using the
102!> Message Passing Interface (MPI)). The model can also be run on vector machines
103!> (less well optimized) and workstations. Versions for the different types of
104!> machines are controlled via cpp-directives.
105!> Model runs are only feasible using the ksh-script mrun.
[1]106!------------------------------------------------------------------------------!
[1682]107 PROGRAM palm
108 
[1]109
[1374]110    USE arrays_3d
[1]111
[1320]112    USE control_parameters,                                                    &
113        ONLY:  coupling_char, coupling_mode, do2d_at_begin, do3d_at_begin,     &
[1666]114               io_blocks, io_group, large_scale_forcing, message_string,       &
115               nudging, simulated_time, simulated_time_chr,                    &
116               user_interface_current_revision,                                &
117               user_interface_required_revision, version, wall_heatflux,       &
118               write_binary
[1320]119
120    USE cpulog,                                                                &
121        ONLY:  cpu_log, log_point, cpu_statistics
122
[1374]123    USE grid_variables,                                                        &
124        ONLY:  fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, wall_e_x, wall_e_y, &
125               wall_u, wall_v, wall_w_x, wall_w_y
126
127    USE indices,                                                               &
[1666]128        ONLY:  ngp_2dh, ngp_2dh_s_inner, nzb_diff_s_inner, nzb_diff_s_outer,   &
129               nzb_diff_u, nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner,  &
130               nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner,             &
131               nzb_w_outer, rflags_invers, rflags_s_inner, wall_flags_0,       &
132               wall_flags_00
[1374]133
[1320]134    USE kinds
135
136    USE ls_forcing_mod,                                                        &
137        ONLY:  init_ls_forcing
138
139    USE nudge_mod,                                                             &
140        ONLY:  init_nudge
141
142    USE particle_attributes,                                                   &
143        ONLY:  particle_advection
144
[1]145    USE pegrid
146
[1374]147    USE statistics,                                                            &
148        ONLY:  hom, rmask, weight_pres, weight_substep
149
[1747]150    USE surface_layer_fluxes_mod,                                              &
151        ONLY:  pt1, qv1, uv_total
152
[1015]153#if defined( __openacc )
154    USE OPENACC
155#endif
156
[1]157    IMPLICIT NONE
158
159!
160!-- Local variables
[1682]161    CHARACTER(LEN=9)  ::  time_to_string  !<
162    CHARACTER(LEN=10) ::  env_string      !< to store string of environment var
163    INTEGER(iwp)      ::  env_stat        !< to hold status of GET_ENV
164    INTEGER(iwp)      ::  i               !<
165    INTEGER(iwp)      ::  myid_openmpi    !< OpenMPI local rank for CUDA aware MPI
[1015]166#if defined( __openacc )
[1682]167    REAL(wp), DIMENSION(100) ::  acc_dum     !<
[1015]168#endif
[1]169
[1539]170    version = 'PALM 4.0'
[1666]171    user_interface_required_revision = 'r1663'
[75]172
[1]173#if defined( __parallel )
174!
175!-- MPI initialisation. comm2d is preliminary set, because
176!-- it will be defined in init_pegrid but is used before in cpu_log.
177    CALL MPI_INIT( ierr )
178    CALL MPI_COMM_SIZE( MPI_COMM_WORLD, numprocs, ierr )
[206]179    CALL MPI_COMM_RANK( MPI_COMM_WORLD, myid, ierr )
[1]180    comm_palm = MPI_COMM_WORLD
181    comm2d    = MPI_COMM_WORLD
182
183!
[206]184!-- Initialize PE topology in case of coupled runs
185    CALL init_coupling
[102]186#endif
187
[1015]188#if defined( __openacc )
[102]189!
[1482]190!-- Get the local MPI rank in case of CUDA aware OpenMPI. Important, if there
191!-- is more than one accelerator board on the node
192    CALL GET_ENVIRONMENT_VARIABLE('OMPI_COMM_WORLD_LOCAL_RANK',                &
193         VALUE=env_string, STATUS=env_stat )
194    READ( env_string, '(I1)' )  myid_openmpi
195    PRINT*, '### local_rank = ', myid_openmpi, '  status=',env_stat
196!
[1015]197!-- Get the number of accelerator boards per node and assign the MPI processes
198!-- to these boards
[1092]199    PRINT*, '*** ACC_DEVICE_NVIDIA = ', ACC_DEVICE_NVIDIA
[1015]200    num_acc_per_node  = ACC_GET_NUM_DEVICES( ACC_DEVICE_NVIDIA )
[1092]201    IF ( numprocs == 1  .AND.  num_acc_per_node > 0 )  num_acc_per_node = 1
[1482]202    PRINT*, '*** myid = ', myid_openmpi, ' num_acc_per_node = ', num_acc_per_node
203    acc_rank = MOD( myid_openmpi, num_acc_per_node )
[1015]204    CALL ACC_SET_DEVICE_NUM ( acc_rank, ACC_DEVICE_NVIDIA )
205!
206!-- Test output (to be removed later)
[1482]207    WRITE (*,'(A,I6,A,I3,A,I3,A,I3)') '*** Connect MPI-Task ', myid_openmpi,   &
208                                      ' to CPU ', acc_rank, ' Devices: ',      &
209                                      num_acc_per_node, ' connected to:',      &
[1015]210                                      ACC_GET_DEVICE_NUM( ACC_DEVICE_NVIDIA )
211#endif
[1092]212
[1015]213!
214!-- Ensure that OpenACC first attaches the GPU devices by copying a dummy data
215!-- region
216    !$acc data copyin( acc_dum )
217
218!
[1]219!-- Initialize measuring of the CPU-time remaining to the run
220    CALL local_tremain_ini
221
222!
223!-- Start of total CPU time measuring.
224    CALL cpu_log( log_point(1), 'total', 'start' )
225    CALL cpu_log( log_point(2), 'initialisation', 'start' )
226
227!
[206]228!-- Open a file for debug output
[1468]229    WRITE (myid_char,'(''_'',I6.6)')  myid
[206]230    OPEN( 9, FILE='DEBUG'//TRIM( coupling_char )//myid_char, FORM='FORMATTED' )
231
232!
[1]233!-- Initialize dvrp logging. Also, one PE maybe split from the global
234!-- communicator for doing the dvrp output. In that case, the number of
235!-- PEs available for PALM is reduced by one and communicator comm_palm
236!-- is changed respectively.
237#if defined( __parallel )
238    CALL MPI_COMM_RANK( comm_palm, myid, ierr )
239!
[102]240!-- TEST OUTPUT (TO BE REMOVED)
241    WRITE(9,*) '*** coupling_mode = "', TRIM( coupling_mode ), '"'
242    CALL LOCAL_FLUSH( 9 )
[215]243    IF ( TRIM( coupling_mode ) /= 'uncoupled' )  THEN
244       PRINT*, '*** PE', myid, ' Global target PE:', target_id, &
245               TRIM( coupling_mode )
246    ENDIF
[102]247#endif
248
[108]249    CALL init_dvrp_logging
250
[102]251!
[108]252!-- Read control parameters from NAMELIST files and read environment-variables
253    CALL parin
254
255!
[1666]256!-- Check for the user's interface version
257    IF ( user_interface_current_revision /= user_interface_required_revision )  &
258    THEN
259       message_string = 'current user-interface revision "' //                  &
260                        TRIM( user_interface_current_revision ) // '" does ' // &
261                        'not match the required revision ' //                   &
262                        TRIM( user_interface_required_revision )
[1668]263        CALL message( 'palm', 'PA0169', 1, 2, 0, 6, 0 )
[1666]264    ENDIF
265
266!
[108]267!-- Determine processor topology and local array indices
268    CALL init_pegrid
269
270!
[1]271!-- Generate grid parameters
272    CALL init_grid
273
274!
[1241]275!-- Initialize nudging if required
276    IF ( nudging )  THEN
277       CALL init_nudge
278    ENDIF
279
280!
281!-- Initialize reading of large scale forcing from external file - if required
282    IF ( large_scale_forcing )  THEN
283       CALL init_ls_forcing
284    ENDIF
285
286!
[1]287!-- Check control parameters and deduce further quantities
288    CALL check_parameters
289
290!
291!-- Initialize all necessary variables
292    CALL init_3d_model
293
294!
295!-- Output of program header
296    IF ( myid == 0 )  CALL header
297
298    CALL cpu_log( log_point(2), 'initialisation', 'stop' )
299
300!
301!-- Set start time in format hh:mm:ss
302    simulated_time_chr = time_to_string( simulated_time )
303
304!
305!-- If required, output of initial arrays
306    IF ( do2d_at_begin )  THEN
307       CALL data_output_2d( 'xy', 0 )
308       CALL data_output_2d( 'xz', 0 )
309       CALL data_output_2d( 'yz', 0 )
310    ENDIF
311    IF ( do3d_at_begin )  THEN
312       CALL data_output_3d( 0 )
313    ENDIF
314
315!
[1015]316!-- Declare and initialize variables in the accelerator memory with their
317!-- host values
[1113]318    !$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]319    !$acc       copyin( tri, tric, dzu, ddzu, ddzw, dd2zu, l_grid, l_wall, ptdf_x, ptdf_y, pt_init, rdf, rdf_sc, ref_state, ug, u_init, vg, v_init, zu, zw )   &
[1747]320    !$acc       copyin( hom, ol, pt1, qs, qsws, qswst, qv1, rif_wall, shf, ts, tswst, us, usws, uswst, uv_total, vsws, vswst, z0, z0h )      &
[1015]321    !$acc       copyin( fxm, fxp, fym, fyp, fwxm, fwxp, fwym, fwyp, nzb_diff_s_inner, nzb_diff_s_outer, nzb_diff_u )       &
322    !$acc       copyin( nzb_diff_v, nzb_s_inner, nzb_s_outer, nzb_u_inner )    &
323    !$acc       copyin( nzb_u_outer, nzb_v_inner, nzb_v_outer, nzb_w_inner )   &
[1221]324    !$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 )  &
325    !$acc       copyin( ngp_2dh, ngp_2dh_s_inner )  &
[1113]326    !$acc       copyin( weight_pres, weight_substep )
[1015]327!
[495]328!-- Integration of the model equations using timestep-scheme
[1]329    CALL time_integration
330
331!
[495]332!-- If required, write binary data for restart runs
333    IF ( write_binary(1:4) == 'true' )  THEN
[759]334
335       CALL cpu_log( log_point(22), 'write_3d_binary', 'start' )
336
[1402]337       CALL location_message( 'writing restart data', .FALSE. )
338
[759]339       CALL check_open( 14 )
340
341       DO  i = 0, io_blocks-1
342          IF ( i == io_group )  THEN
[1]343!
[759]344!--          Write flow field data
345             CALL write_3d_binary
346          ENDIF
347#if defined( __parallel )
348          CALL MPI_BARRIER( comm2d, ierr )
349#endif
350       ENDDO
351
[1402]352       CALL location_message( 'finished', .TRUE. )
353
[759]354       CALL cpu_log( log_point(22), 'write_3d_binary', 'stop' )
355
[495]356!
357!--    If required, write particle data
[849]358       IF ( particle_advection )  CALL lpm_write_restart_file
[495]359    ENDIF
[1]360
361!
362!-- If required, repeat output of header including the required CPU-time
363    IF ( myid == 0 )  CALL header
364!
365!-- If required, final user-defined actions, and
366!-- last actions on the open files and close files. Unit 14 was opened
367!-- in write_3d_binary but it is closed here, to allow writing on this
368!-- unit in routine user_last_actions.
369    CALL cpu_log( log_point(4), 'last actions', 'start' )
[759]370    DO  i = 0, io_blocks-1
371       IF ( i == io_group )  THEN
372          CALL user_last_actions
373          IF ( write_binary(1:4) == 'true' )  CALL close_file( 14 )
374       ENDIF
375#if defined( __parallel )
376       CALL MPI_BARRIER( comm2d, ierr )
377#endif
378    ENDDO
[1]379    CALL close_file( 0 )
380    CALL close_dvrp
381    CALL cpu_log( log_point(4), 'last actions', 'stop' )
382
[102]383#if defined( __mpi2 )
[1]384!
[206]385!-- Test exchange via intercommunicator in case of a MPI-2 coupling
[102]386    IF ( coupling_mode == 'atmosphere_to_ocean' )  THEN
387       i = 12345 + myid
388       CALL MPI_SEND( i, 1, MPI_INTEGER, myid, 11, comm_inter, ierr )
389    ELSEIF ( coupling_mode == 'ocean_to_atmosphere' )  THEN
390       CALL MPI_RECV( i, 1, MPI_INTEGER, myid, 11, comm_inter, status, ierr )
391       PRINT*, '### myid: ', myid, '   received from atmosphere:  i = ', i
392    ENDIF
393#endif
394
395!
[1015]396!-- Close the OpenACC dummy data region
397    !$acc end data
398    !$acc end data
399
400!
[1]401!-- Take final CPU-time for CPU-time analysis
402    CALL cpu_log( log_point(1), 'total', 'stop' )
403    CALL cpu_statistics
404
405#if defined( __parallel )
406    CALL MPI_FINALIZE( ierr )
407#endif
408
409 END PROGRAM palm
Note: See TracBrowser for help on using the repository browser.