[92fab65] | 1 | ! SPDX-FileCopyrightText: FLEXPART 1998-2019, see flexpart_license.txt |
---|
| 2 | ! SPDX-License-Identifier: GPL-3.0-or-later |
---|
[332fbbd] | 3 | |
---|
[8a65cb0] | 4 | module mpi_mod |
---|
| 5 | |
---|
| 6 | !***************************************************************************** |
---|
| 7 | ! * |
---|
| 8 | ! DESCRIPTION * |
---|
| 9 | ! This module contains subroutines and common variables used for the * |
---|
| 10 | ! MPI parallelization of FLEXPART. * |
---|
| 11 | ! * |
---|
| 12 | ! NOTE * |
---|
| 13 | ! Depending on the MPI library installed on your system (e.g. mpich2, * |
---|
| 14 | ! OpenMPI) you may need to choose below in this file between * |
---|
| 15 | ! use mpi * |
---|
| 16 | ! (if the MPI library comes with the file 'mpi.mod'); or * |
---|
| 17 | ! include 'mpif.h' * |
---|
| 18 | ! * |
---|
| 19 | ! * |
---|
| 20 | !***************************************************************************** |
---|
| 21 | ! * |
---|
| 22 | ! Variables: * |
---|
| 23 | ! * |
---|
| 24 | ! mp_ierr MPI error code * |
---|
| 25 | ! mp_np Number of MPI processes * |
---|
| 26 | ! mp_pid Process ID of each MPI process * |
---|
| 27 | ! mp_seed Parameter for random number seed * |
---|
[5f9d14a] | 28 | ! read_grp_min Minimum number of processes at which one will be * |
---|
| 29 | ! used as reader * |
---|
[8a65cb0] | 30 | ! numpart_mpi, Number of particles per node * |
---|
| 31 | ! maxpart_mpi * |
---|
| 32 | ! mp_partid MPI process ID for particle calculation * |
---|
| 33 | ! mp_partgroup_ Refers to the subset of processors performing * |
---|
| 34 | ! loops over particles. Will be all processes * |
---|
| 35 | ! unless a dedicated process runs getfields/readwind * |
---|
[5f9d14a] | 36 | ! lmp_sync If .false., use asynchronous MPI * |
---|
[6a678e3] | 37 | ! mp_cp Real precision to use for deposition fields * |
---|
[8a65cb0] | 38 | ! * |
---|
| 39 | ! * |
---|
| 40 | ! * |
---|
| 41 | ! * |
---|
| 42 | !***************************************************************************** |
---|
[6ecb30a] | 43 | ! * |
---|
| 44 | ! Modification by DJM, 2017-05-09 - added #ifdef USE_MPIINPLACE cpp * |
---|
| 45 | ! directive to mpif_tm_reduce_grid() to insure that MPI_IN_PLACE is * |
---|
| 46 | ! used in the MPI_Reduce() only if specifically compiled with that * |
---|
| 47 | ! directive. * |
---|
| 48 | ! * |
---|
| 49 | !***************************************************************************** |
---|
[8a65cb0] | 50 | |
---|
| 51 | use mpi |
---|
[1c3c778] | 52 | use par_mod, only: dp,sp |
---|
[8a65cb0] | 53 | use com_mod, only: lroot |
---|
| 54 | |
---|
| 55 | implicit none |
---|
| 56 | |
---|
| 57 | ! include 'mpif.h' |
---|
| 58 | |
---|
| 59 | public |
---|
| 60 | |
---|
| 61 | ! Set aside a process for reading windfields if using at least these many processes |
---|
| 62 | !================================================== |
---|
[5f9d14a] | 63 | integer, parameter, private :: read_grp_min=4 |
---|
[8a65cb0] | 64 | !================================================== |
---|
| 65 | |
---|
| 66 | ! Variables for each MPI process in the world group |
---|
| 67 | integer :: mp_ierr, mp_np, mp_pid, mp_partid |
---|
| 68 | integer, private :: world_group_id |
---|
| 69 | |
---|
| 70 | ! Variables for MPI processes in the 'particle' group |
---|
| 71 | integer, allocatable, dimension(:) :: mp_partgroup_rank |
---|
[0a94e13] | 72 | integer, allocatable, dimension(:) :: npart_per_process |
---|
[8a65cb0] | 73 | integer :: mp_partgroup_comm, mp_partgroup_pid, mp_partgroup_np |
---|
| 74 | |
---|
| 75 | integer :: mp_seed=0 |
---|
| 76 | integer, parameter :: mp_sp=MPI_REAL4, mp_dp=MPI_REAL8 |
---|
[6a678e3] | 77 | integer :: mp_cp |
---|
[8a65cb0] | 78 | integer, parameter :: id_root=0 ! master process |
---|
| 79 | |
---|
| 80 | ! MPI tags/requests for send/receive operation |
---|
[5f9d14a] | 81 | integer :: tm1 |
---|
[861805a] | 82 | integer, parameter :: nvar_async=26 |
---|
[78e62dc] | 83 | !integer, dimension(:), allocatable :: tags |
---|
[8a65cb0] | 84 | integer, dimension(:), allocatable :: reqs |
---|
| 85 | |
---|
[861805a] | 86 | ! Status array used for certain MPI operations (MPI_RECV) |
---|
| 87 | integer, dimension(MPI_STATUS_SIZE) :: mp_status |
---|
| 88 | |
---|
[8a65cb0] | 89 | |
---|
| 90 | integer :: id_read ! readwind/getfield process |
---|
| 91 | integer :: numpart_mpi,maxpart_mpi ! number of particles per node |
---|
| 92 | integer :: tot_numpart=0 |
---|
| 93 | integer :: mp_comm_used ! global or subgroup communicator |
---|
| 94 | |
---|
| 95 | logical :: lmpreader=.false. ! is set to true for reading process(es) only. |
---|
| 96 | logical :: lmp_use_reader=.false. ! true if separate readwind process is used |
---|
| 97 | |
---|
[62e65c7] | 98 | ! .true. if only using synchronous MPI send/recv (default) |
---|
[8a65cb0] | 99 | ! If setting this to .false., numwfmem must be set to 3 |
---|
| 100 | !=============================================================================== |
---|
[a1f4dd6] | 101 | logical :: lmp_sync=.true. |
---|
[8a65cb0] | 102 | !=============================================================================== |
---|
| 103 | |
---|
[5f9d14a] | 104 | ! mp_dbg_mode Used for debugging MPI. |
---|
[8a65cb0] | 105 | ! mp_dev_mode various checks related to debugging the parallel code |
---|
| 106 | ! mp_dbg_out write some arrays to data file for debugging |
---|
| 107 | ! mp_measure_time Measure cpu/wall time, write out at end of run |
---|
| 108 | ! mp_time_barrier Measure MPI barrier time |
---|
[f55fdce] | 109 | ! mp_exact_numpart Use an extra MPI communication to give the exact number of particles |
---|
[0a94e13] | 110 | ! to standard output (this does not otherwise affect the simulation) |
---|
[8a65cb0] | 111 | logical, parameter :: mp_dbg_mode = .false. |
---|
| 112 | logical, parameter :: mp_dev_mode = .false. |
---|
| 113 | logical, parameter :: mp_dbg_out = .false. |
---|
[16b61a5] | 114 | logical, parameter :: mp_time_barrier=.true. |
---|
[ec7fc72] | 115 | logical, parameter :: mp_measure_time=.false. |
---|
[f55fdce] | 116 | logical, parameter :: mp_exact_numpart=.true. |
---|
[8a65cb0] | 117 | |
---|
| 118 | ! for measuring CPU/Wall time |
---|
[38b7917] | 119 | real(sp),private :: mp_comm_time_beg, mp_comm_time_end, mp_comm_time_total=0. |
---|
| 120 | real(dp),private :: mp_comm_wtime_beg, mp_comm_wtime_end, mp_comm_wtime_total=0. |
---|
| 121 | real(sp),private :: mp_root_time_beg, mp_root_time_end, mp_root_time_total=0. |
---|
| 122 | real(dp),private :: mp_root_wtime_beg, mp_root_wtime_end, mp_root_wtime_total=0. |
---|
| 123 | real(sp),private :: mp_barrier_time_beg, mp_barrier_time_end, mp_barrier_time_total=0. |
---|
| 124 | real(dp),private :: mp_barrier_wtime_beg, mp_barrier_wtime_end, mp_barrier_wtime_total=0. |
---|
| 125 | real(sp),private :: tm_nploop_beg, tm_nploop_end, tm_nploop_total=0. |
---|
| 126 | real(sp),private :: tm_tot_beg, tm_tot_end, tm_tot_total=0. |
---|
| 127 | real(dp),private :: mp_getfields_wtime_beg, mp_getfields_wtime_end, mp_getfields_wtime_total=0. |
---|
| 128 | real(sp),private :: mp_getfields_time_beg, mp_getfields_time_end, mp_getfields_time_total=0. |
---|
[0a94e13] | 129 | real(dp),private :: mp_readwind_wtime_beg, mp_readwind_wtime_end, mp_readwind_wtime_total=0. |
---|
| 130 | real(sp),private :: mp_readwind_time_beg, mp_readwind_time_end, mp_readwind_time_total=0. |
---|
[38b7917] | 131 | real(dp),private :: mp_io_wtime_beg, mp_io_wtime_end, mp_io_wtime_total=0. |
---|
| 132 | real(sp),private :: mp_io_time_beg, mp_io_time_end, mp_io_time_total=0. |
---|
| 133 | real(dp),private :: mp_wetdepo_wtime_beg, mp_wetdepo_wtime_end, mp_wetdepo_wtime_total=0. |
---|
| 134 | real(sp),private :: mp_wetdepo_time_beg, mp_wetdepo_time_end, mp_wetdepo_time_total=0. |
---|
| 135 | real(dp),private :: mp_advance_wtime_beg, mp_advance_wtime_end, mp_advance_wtime_total=0. |
---|
| 136 | real(dp),private :: mp_conccalc_time_beg, mp_conccalc_time_end, mp_conccalc_time_total=0. |
---|
| 137 | real(dp),private :: mp_total_wtime_beg, mp_total_wtime_end, mp_total_wtime_total=0. |
---|
| 138 | real(dp),private :: mp_vt_wtime_beg, mp_vt_wtime_end, mp_vt_wtime_total |
---|
| 139 | real(sp),private :: mp_vt_time_beg, mp_vt_time_end, mp_vt_time_total |
---|
[8a65cb0] | 140 | |
---|
| 141 | ! dat_lun logical unit number for i/o |
---|
| 142 | integer, private :: dat_lun |
---|
| 143 | |
---|
[861805a] | 144 | ! Temporary arrays for particles (allocated and deallocated as needed) |
---|
| 145 | integer, allocatable, dimension(:) :: nclass_tmp, npoint_tmp, itra1_tmp, idt_tmp, & |
---|
| 146 | & itramem_tmp, itrasplit_tmp |
---|
| 147 | real(kind=dp), allocatable, dimension(:) :: xtra1_tmp, ytra1_tmp |
---|
| 148 | real, allocatable, dimension(:) :: ztra1_tmp |
---|
| 149 | real, allocatable, dimension(:,:) :: xmass1_tmp |
---|
| 150 | |
---|
| 151 | ! mp_redist_fract Exchange particles between processes if relative difference |
---|
| 152 | ! is larger. A good value is between 0.0 and 0.5 |
---|
| 153 | ! mp_maxpart_factor Allocate more memory per process than strictly needed |
---|
| 154 | ! (safety factor). Recommended value between 1.5 and 2.5 |
---|
| 155 | ! mp_min_redist Do not redistribute particles if below this limit |
---|
| 156 | real, parameter :: mp_redist_fract=0.2, mp_maxpart_factor=1.5 |
---|
| 157 | integer,parameter :: mp_min_redist=100000 |
---|
| 158 | |
---|
| 159 | |
---|
[8a65cb0] | 160 | contains |
---|
| 161 | |
---|
| 162 | subroutine mpif_init |
---|
| 163 | !*********************************************************************** |
---|
| 164 | ! DESCRIPTION |
---|
| 165 | ! Initialize MPI. |
---|
| 166 | ! |
---|
| 167 | ! Create the global communicator MPI_COMM_WORLD |
---|
| 168 | ! If using a separate MPI process for getfields/readwind, a subgroup |
---|
| 169 | ! is created for the other processes. |
---|
| 170 | ! |
---|
| 171 | ! VARIABLES |
---|
| 172 | ! mpi_mode default 0, set to 2/3 if running MPI version |
---|
| 173 | ! mp_np number of running processes, decided at run-time |
---|
| 174 | !*********************************************************************** |
---|
[328fdf9] | 175 | use par_mod, only: maxpart, numwfmem, dep_prec, maxreceptor, maxspec |
---|
| 176 | use com_mod, only: mpi_mode, verbosity, creceptor0 |
---|
[8a65cb0] | 177 | |
---|
| 178 | implicit none |
---|
| 179 | |
---|
| 180 | integer :: i,j,s,addmaxpart=0 |
---|
| 181 | |
---|
[6a678e3] | 182 | ! Each process gets an ID (mp_pid) in the range 0,..,mp_np-1 |
---|
[8a65cb0] | 183 | call MPI_INIT(mp_ierr) |
---|
| 184 | if (mp_ierr /= 0) goto 100 |
---|
| 185 | call MPI_COMM_RANK(MPI_COMM_WORLD, mp_pid, mp_ierr) |
---|
| 186 | if (mp_ierr /= 0) goto 100 |
---|
| 187 | call MPI_COMM_SIZE(MPI_COMM_WORLD, mp_np, mp_ierr) |
---|
| 188 | if (mp_ierr /= 0) goto 100 |
---|
| 189 | |
---|
| 190 | |
---|
[6a678e3] | 191 | ! Variable mpi_mode is used to handle subroutines common to parallel/serial version |
---|
[8a65cb0] | 192 | if (lmp_sync) then |
---|
| 193 | mpi_mode=2 ! hold 2 windfields in memory |
---|
| 194 | else |
---|
| 195 | mpi_mode=3 ! hold 3 windfields in memory |
---|
| 196 | end if |
---|
| 197 | |
---|
| 198 | if (mp_pid.ne.0) then |
---|
| 199 | lroot = .false. |
---|
| 200 | end if |
---|
| 201 | |
---|
[6a678e3] | 202 | ! Set MPI precision to use for transferring deposition fields |
---|
| 203 | !************************************************************ |
---|
| 204 | if (dep_prec==dp) then |
---|
| 205 | mp_cp = MPI_REAL8 |
---|
[861805a] | 206 | ! TODO: write info message for serial version as well |
---|
[8ed5f11] | 207 | if (lroot.and.verbosity>0) write(*,*) 'Using double precision for deposition fields' |
---|
[6a678e3] | 208 | else if (dep_prec==sp) then |
---|
| 209 | mp_cp = MPI_REAL4 |
---|
[8ed5f11] | 210 | if (lroot.and.verbosity>0) write(*,*) 'Using single precision for deposition fields' |
---|
[6a678e3] | 211 | else |
---|
| 212 | write(*,*) 'ERROR: something went wrong setting MPI real precision' |
---|
| 213 | stop |
---|
| 214 | end if |
---|
| 215 | |
---|
[8a65cb0] | 216 | ! Check for sensible combination of parameters |
---|
| 217 | !********************************************* |
---|
| 218 | |
---|
[62e65c7] | 219 | if (.not.lmp_sync.and.numwfmem.ne.3) then |
---|
| 220 | if (lroot) then |
---|
| 221 | write(*,FMT='(80("#"))') |
---|
| 222 | write(*,*) '#### mpi_mod::mpif_init> ERROR: ', & |
---|
| 223 | & 'numwfmem must be set to 3 for asyncronous reading ####' |
---|
| 224 | write(*,FMT='(80("#"))') |
---|
| 225 | end if |
---|
| 226 | call MPI_FINALIZE(mp_ierr) |
---|
[8a65cb0] | 227 | stop |
---|
| 228 | else if (lmp_sync.and.numwfmem.ne.2.and.lroot) then |
---|
| 229 | write(*,FMT='(80("#"))') |
---|
| 230 | write(*,*) '#### mpi_mod::mpif_init> WARNING: ', & |
---|
[a1f4dd6] | 231 | & 'numwfmem should be set to 2 for syncronous' |
---|
[62e65c7] | 232 | write(*,*) ' reading. Results will still be valid, but unneccesary memory & |
---|
| 233 | &is allocated.' |
---|
[8a65cb0] | 234 | write(*,FMT='(80("#"))') |
---|
[5f9d14a] | 235 | ! Force "syncronized" version if all processes will call getfields |
---|
[d6a0977] | 236 | else if ((.not.lmp_sync.and.mp_np.lt.read_grp_min).or.(mp_np.eq.1)) then |
---|
[5f9d14a] | 237 | if (lroot) then |
---|
| 238 | write(*,FMT='(80("#"))') |
---|
| 239 | write(*,*) '#### mpi_mod::mpif_init> WARNING: ', & |
---|
| 240 | & 'all procs call getfields. Setting lmp_sync=.true.' |
---|
| 241 | write(*,FMT='(80("#"))') |
---|
| 242 | end if |
---|
[16b61a5] | 243 | lmp_sync=.true. |
---|
[8a65cb0] | 244 | end if |
---|
[6b22af9] | 245 | |
---|
| 246 | ! TODO: Add more warnings for unimplemented flexpart features |
---|
[8a65cb0] | 247 | |
---|
| 248 | ! Set ID of process that calls getfield/readwind. |
---|
| 249 | ! Using the last process in the group ensures statistical identical results |
---|
| 250 | ! as running with one process less but not using separate read process |
---|
| 251 | !********************************************************************** |
---|
| 252 | |
---|
[16b61a5] | 253 | ! id_read = min(mp_np-1, 1) |
---|
[8a65cb0] | 254 | id_read = mp_np-1 |
---|
| 255 | |
---|
| 256 | if (mp_pid.eq.id_read) lmpreader=.true. |
---|
| 257 | |
---|
| 258 | call MPI_Comm_group (MPI_COMM_WORLD, world_group_id, mp_ierr) |
---|
| 259 | |
---|
[5f9d14a] | 260 | ! Create a MPI group/communicator that will calculate trajectories. |
---|
[8a65cb0] | 261 | ! Skip this step if program is run with only a few processes |
---|
| 262 | !************************************************************************ |
---|
| 263 | allocate(mp_partgroup_rank(0:mp_np-2)) |
---|
| 264 | |
---|
| 265 | ! This allows checking for allocation of arrays when no subgroup is used |
---|
| 266 | mp_partgroup_pid=0 |
---|
| 267 | |
---|
| 268 | if (read_grp_min.lt.2) then |
---|
| 269 | write(*,*) '#### mpi_mod::mpif_init> ERROR ####', & |
---|
| 270 | & 'read_grp_min should be at least 2. Exiting' |
---|
| 271 | stop |
---|
| 272 | end if |
---|
| 273 | |
---|
| 274 | if (mp_np.ge.read_grp_min) then |
---|
| 275 | lmp_use_reader = .true. |
---|
| 276 | |
---|
[5f9d14a] | 277 | ! Map the subgroup IDs= 0,1,2,...,mp_np-2, skipping reader process |
---|
[8a65cb0] | 278 | j=-1 |
---|
| 279 | do i=0, mp_np-2 !loop over all (subgroup) IDs |
---|
| 280 | j=j+1 |
---|
| 281 | if (i.eq.id_read) then |
---|
| 282 | j=j+1 |
---|
| 283 | end if |
---|
| 284 | mp_partgroup_rank(i) = j |
---|
| 285 | end do |
---|
| 286 | |
---|
| 287 | call MPI_Group_incl (world_group_id, mp_np-1, mp_partgroup_rank, & |
---|
| 288 | &mp_partgroup_pid, mp_ierr) |
---|
| 289 | if (mp_ierr /= 0) goto 100 |
---|
| 290 | call MPI_Comm_create (MPI_COMM_WORLD, mp_partgroup_pid, mp_partgroup_comm, mp_ierr) |
---|
| 291 | if (mp_ierr /= 0) goto 100 |
---|
| 292 | |
---|
| 293 | if (mp_pid.ne.id_read) then |
---|
| 294 | call MPI_Comm_rank (mp_partgroup_comm, mp_partgroup_pid, mp_ierr) |
---|
| 295 | if (mp_ierr /= 0) goto 100 |
---|
| 296 | |
---|
| 297 | ! Get size of new group (=mp_np-1) |
---|
| 298 | call MPI_COMM_SIZE(mp_partgroup_comm, mp_partgroup_np, mp_ierr) |
---|
| 299 | if (mp_ierr /= 0) goto 100 |
---|
| 300 | if (mp_partgroup_np.ne.mp_np-1) then |
---|
| 301 | write(*,*) '#### mpi_mod:: mpif_init> ERROR ####& |
---|
| 302 | & mp_partgroup_np.ne.mp_np-1' |
---|
| 303 | stop |
---|
| 304 | endif |
---|
| 305 | |
---|
| 306 | else |
---|
| 307 | mp_partgroup_pid = -1 |
---|
| 308 | end if |
---|
| 309 | end if |
---|
| 310 | |
---|
| 311 | if (lmp_use_reader) then |
---|
| 312 | mp_comm_used = mp_partgroup_comm |
---|
| 313 | mp_partid = mp_partgroup_pid |
---|
| 314 | mp_partgroup_np=mp_np-1 |
---|
| 315 | else |
---|
| 316 | mp_comm_used = MPI_COMM_WORLD |
---|
| 317 | mp_partgroup_np = mp_np |
---|
| 318 | mp_partid = mp_pid |
---|
| 319 | end if |
---|
| 320 | |
---|
| 321 | ! Set maxpart per process |
---|
[328fdf9] | 322 | ! ESO 08/2016: Increase maxpart per process, in case of unbalanced distribution |
---|
[4c64400] | 323 | maxpart_mpi=int(mp_maxpart_factor*real(maxpart)/real(mp_partgroup_np)) |
---|
[b5127f9] | 324 | if (mp_np == 1) maxpart_mpi = maxpart |
---|
[8a65cb0] | 325 | |
---|
| 326 | ! Do not allocate particle data arrays for readwind process |
---|
| 327 | if (lmpreader.and.lmp_use_reader) then |
---|
| 328 | maxpart_mpi=0 |
---|
| 329 | end if |
---|
| 330 | |
---|
| 331 | ! Set random seed for each non-root process |
---|
| 332 | if (mp_pid.gt.0) then |
---|
| 333 | s = 244 |
---|
| 334 | mp_seed = -abs(mod((s*181)*((mp_pid-83)*359), 104729)) |
---|
| 335 | end if |
---|
[5f9d14a] | 336 | if (mp_dbg_mode) then |
---|
| 337 | mp_seed=0 |
---|
| 338 | if (lroot) write(*,*) 'MPI: setting seed=0' |
---|
| 339 | end if |
---|
[8a65cb0] | 340 | |
---|
| 341 | ! Allocate request array for asynchronous MPI |
---|
| 342 | if (.not.lmp_sync) then |
---|
| 343 | allocate(reqs(0:nvar_async*mp_np-1)) |
---|
[5f9d14a] | 344 | reqs(:)=MPI_REQUEST_NULL |
---|
[8a65cb0] | 345 | else |
---|
| 346 | allocate(reqs(0:1)) |
---|
[5f9d14a] | 347 | reqs(:)=MPI_REQUEST_NULL |
---|
[8a65cb0] | 348 | end if |
---|
| 349 | |
---|
[0a94e13] | 350 | ! Allocate array for number of particles per process |
---|
| 351 | allocate(npart_per_process(0:mp_partgroup_np-1)) |
---|
| 352 | |
---|
[328fdf9] | 353 | ! Write whether MPI_IN_PLACE is used or not |
---|
| 354 | #ifdef USE_MPIINPLACE |
---|
| 355 | if (lroot) write(*,*) 'Using MPI_IN_PLACE operations' |
---|
| 356 | #else |
---|
| 357 | if (lroot) allocate(creceptor0(maxreceptor,maxspec)) |
---|
| 358 | if (lroot) write(*,*) 'Not using MPI_IN_PLACE operations' |
---|
| 359 | #endif |
---|
[8a65cb0] | 360 | goto 101 |
---|
| 361 | |
---|
| 362 | 100 write(*,*) '#### mpi_mod::mpif_init> ERROR ####', mp_ierr |
---|
| 363 | stop |
---|
| 364 | |
---|
| 365 | 101 end subroutine mpif_init |
---|
| 366 | |
---|
| 367 | |
---|
| 368 | subroutine mpif_mpi_barrier |
---|
| 369 | !*********************************************************************** |
---|
| 370 | ! Set MPI barrier (all processes are synchronized here), with the |
---|
| 371 | ! possible exception of a separate 'readwind' process. |
---|
| 372 | ! Optionally measure cpu/wall time. |
---|
| 373 | ! |
---|
| 374 | !*********************************************************************** |
---|
| 375 | implicit none |
---|
| 376 | |
---|
| 377 | if (mp_time_barrier) then |
---|
| 378 | call cpu_time(mp_barrier_time_beg) |
---|
| 379 | mp_barrier_wtime_beg = mpi_wtime() |
---|
| 380 | endif |
---|
| 381 | |
---|
| 382 | call MPI_BARRIER(mp_comm_used, mp_ierr) |
---|
| 383 | if (mp_ierr /= 0) goto 100 |
---|
| 384 | |
---|
| 385 | if (mp_time_barrier) then |
---|
| 386 | call cpu_time(mp_barrier_time_end) |
---|
| 387 | mp_barrier_wtime_end = mpi_wtime() |
---|
| 388 | |
---|
| 389 | mp_barrier_time_total=mp_barrier_time_total+& |
---|
| 390 | &(mp_barrier_time_end-mp_barrier_time_beg) |
---|
| 391 | mp_barrier_wtime_total=mp_barrier_wtime_total+& |
---|
| 392 | &(mp_barrier_wtime_end-mp_barrier_wtime_beg) |
---|
| 393 | end if |
---|
| 394 | |
---|
| 395 | goto 101 |
---|
| 396 | |
---|
| 397 | 100 write(*,*) '#### mpi_mod::mpif_mpi_barrier> ERROR ####', mp_ierr |
---|
| 398 | stop |
---|
| 399 | |
---|
| 400 | 101 end subroutine mpif_mpi_barrier |
---|
| 401 | |
---|
| 402 | |
---|
| 403 | subroutine mpif_com_mod_allocate |
---|
| 404 | !******************************************************************************* |
---|
| 405 | ! Dynamic allocation of arrays (in serial code these have size maxpart) |
---|
| 406 | ! |
---|
[32b49c3] | 407 | ! Not in use anymore, moved to com_mod for interoperability with serial version |
---|
| 408 | ! |
---|
| 409 | ! TODO: remove |
---|
[8a65cb0] | 410 | !******************************************************************************* |
---|
| 411 | use com_mod |
---|
| 412 | |
---|
| 413 | implicit none |
---|
| 414 | |
---|
| 415 | integer :: array_size |
---|
| 416 | |
---|
| 417 | array_size = maxpart_mpi |
---|
| 418 | |
---|
| 419 | allocate(itra1(array_size),npoint(array_size),& |
---|
| 420 | & nclass(array_size),idt(array_size),& |
---|
| 421 | & itramem(array_size),itrasplit(array_size),& |
---|
| 422 | & xtra1(array_size),ytra1(array_size),& |
---|
| 423 | & ztra1(array_size),xmass1(array_size, maxspec)) |
---|
| 424 | |
---|
| 425 | allocate(uap(array_size),ucp(array_size),& |
---|
| 426 | & uzp(array_size),us(array_size),& |
---|
| 427 | & vs(array_size),& |
---|
| 428 | & ws(array_size),cbt(array_size)) |
---|
| 429 | |
---|
| 430 | |
---|
| 431 | end subroutine mpif_com_mod_allocate |
---|
| 432 | |
---|
| 433 | |
---|
| 434 | subroutine mpif_tm_send_dims |
---|
| 435 | !*********************************************************************** |
---|
| 436 | ! Distribute array dimensions from pid0 to all processes. |
---|
| 437 | ! numpart_mpi/numpart is sent to allow dynamic allocation |
---|
| 438 | ! |
---|
| 439 | ! Currently not in use. |
---|
| 440 | ! |
---|
| 441 | ! Variables of similar type (integer, real) are placed in an array |
---|
| 442 | ! and sent collectively, to avoid the overhead associated with individual |
---|
| 443 | ! MPI send/receives |
---|
| 444 | ! |
---|
| 445 | ! |
---|
| 446 | !*********************************************************************** |
---|
| 447 | use com_mod |
---|
| 448 | |
---|
| 449 | implicit none |
---|
| 450 | |
---|
| 451 | integer :: add_part=0 |
---|
| 452 | |
---|
[5f9d14a] | 453 | call MPI_Bcast(numpart, 1, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 454 | |
---|
| 455 | ! MPI subgroup does the particle-loop |
---|
| 456 | if (lmp_use_reader) then |
---|
| 457 | if (mod(numpart,mp_partgroup_np).ne.0) add_part=1 |
---|
| 458 | numpart_mpi=int(numpart/mp_partgroup_np)+add_part |
---|
| 459 | else |
---|
| 460 | if (mod(numpart,mp_np).ne.0) add_part=1 |
---|
| 461 | numpart_mpi=int(numpart/mp_np)+add_part |
---|
| 462 | end if |
---|
| 463 | |
---|
| 464 | |
---|
| 465 | ! redefine numpart as 'numpart per process' throughout the code |
---|
| 466 | !************************************************************** |
---|
| 467 | numpart = numpart_mpi |
---|
| 468 | |
---|
| 469 | end subroutine mpif_tm_send_dims |
---|
| 470 | |
---|
| 471 | |
---|
[861805a] | 472 | subroutine mpif_send_part_properties(num_part) |
---|
| 473 | !*********************************************************************** |
---|
| 474 | ! Distribute particle properties from root to all processes. |
---|
| 475 | ! |
---|
| 476 | ! Used by init_domainfill_mpi |
---|
| 477 | ! |
---|
| 478 | ! Variables: |
---|
| 479 | ! |
---|
| 480 | ! num_part input, number of particles per process (rounded up) |
---|
| 481 | ! |
---|
| 482 | !*********************************************************************** |
---|
| 483 | use com_mod |
---|
| 484 | |
---|
| 485 | implicit none |
---|
| 486 | |
---|
| 487 | integer,intent(in) :: num_part |
---|
| 488 | integer :: i,jj, addone |
---|
| 489 | |
---|
[16b61a5] | 490 | ! Exit if too many particles |
---|
| 491 | if (num_part.gt.maxpart_mpi) then |
---|
| 492 | write(*,*) '#####################################################' |
---|
| 493 | write(*,*) '#### ERROR - TOTAL NUMBER OF PARTICLES REQUIRED ####' |
---|
| 494 | write(*,*) '#### EXCEEDS THE MAXIMUM ALLOWED NUMBER. REDUCE ####' |
---|
| 495 | write(*,*) '#### EITHER NUMBER OF PARTICLES PER RELEASE POINT####' |
---|
| 496 | write(*,*) '#### OR INCREASE MAXPART. ####' |
---|
| 497 | write(*,*) '#####################################################' |
---|
| 498 | ! call MPI_FINALIZE(mp_ierr) |
---|
| 499 | stop |
---|
| 500 | end if |
---|
| 501 | |
---|
| 502 | |
---|
[861805a] | 503 | ! Time for MPI communications |
---|
| 504 | !**************************** |
---|
| 505 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 506 | |
---|
| 507 | ! Distribute variables, send from pid 0 to other processes (including itself) |
---|
| 508 | !**************************************************************************** |
---|
| 509 | |
---|
| 510 | call MPI_SCATTER(nclass_tmp,num_part,MPI_INTEGER,nclass,& |
---|
| 511 | &num_part,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
| 512 | if (mp_ierr /= 0) goto 600 |
---|
| 513 | call MPI_SCATTER(npoint_tmp,num_part,MPI_INTEGER,npoint,& |
---|
| 514 | &num_part,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
| 515 | if (mp_ierr /= 0) goto 600 |
---|
| 516 | call MPI_SCATTER(itra1_tmp,num_part,MPI_INTEGER,itra1,& |
---|
| 517 | &num_part,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
| 518 | if (mp_ierr /= 0) goto 600 |
---|
| 519 | call MPI_SCATTER(idt_tmp,num_part,MPI_INTEGER,idt,& |
---|
| 520 | &num_part,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
| 521 | if (mp_ierr /= 0) goto 600 |
---|
| 522 | call MPI_SCATTER(itramem_tmp,num_part,MPI_INTEGER,itramem,& |
---|
| 523 | &num_part,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
| 524 | if (mp_ierr /= 0) goto 600 |
---|
| 525 | call MPI_SCATTER(itrasplit_tmp,num_part,MPI_INTEGER,itrasplit,& |
---|
| 526 | &num_part,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
| 527 | if (mp_ierr /= 0) goto 600 |
---|
| 528 | call MPI_SCATTER(xtra1_tmp,num_part,mp_dp,xtra1,& |
---|
| 529 | &num_part,mp_dp,id_root,mp_comm_used,mp_ierr) |
---|
| 530 | if (mp_ierr /= 0) goto 600 |
---|
| 531 | call MPI_SCATTER(ytra1_tmp,num_part,mp_dp,ytra1,& |
---|
| 532 | &num_part,mp_dp,id_root,mp_comm_used,mp_ierr) |
---|
| 533 | if (mp_ierr /= 0) goto 600 |
---|
| 534 | call MPI_SCATTER(ztra1_tmp,num_part,mp_sp,ztra1,& |
---|
| 535 | &num_part,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
| 536 | if (mp_ierr /= 0) goto 600 |
---|
| 537 | do i=1,nspec |
---|
| 538 | call MPI_SCATTER(xmass1_tmp(:,i),num_part,mp_sp,xmass1(:,i),& |
---|
| 539 | &num_part,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
| 540 | if (mp_ierr /= 0) goto 600 |
---|
| 541 | end do |
---|
| 542 | |
---|
| 543 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 544 | |
---|
| 545 | goto 601 |
---|
| 546 | |
---|
| 547 | 600 write(*,*) "mpi_mod> mp_ierr \= 0", mp_ierr |
---|
| 548 | stop |
---|
| 549 | |
---|
| 550 | ! After the transfer of particles it it possible that the value of |
---|
[16b61a5] | 551 | ! "numpart" is larger than the actual used, so we reduce it if there are |
---|
[861805a] | 552 | ! invalid particles at the end of the arrays |
---|
[b5127f9] | 553 | |
---|
[328fdf9] | 554 | 601 do i=numpart, 1, -1 |
---|
[861805a] | 555 | if (itra1(i).eq.-999999999) then |
---|
| 556 | numpart=numpart-1 |
---|
| 557 | else |
---|
| 558 | exit |
---|
| 559 | end if |
---|
| 560 | end do |
---|
| 561 | |
---|
| 562 | |
---|
| 563 | end subroutine mpif_send_part_properties |
---|
| 564 | |
---|
| 565 | |
---|
| 566 | subroutine mpif_calculate_part_redist(itime) |
---|
| 567 | !*********************************************************************** |
---|
| 568 | ! Calculate number of particles to redistribute between processes. This routine |
---|
| 569 | ! can be called at regular time intervals to keep a level number of |
---|
| 570 | ! particles on each process. |
---|
| 571 | ! |
---|
| 572 | ! First, all processes report their local "numpart" to each other, which is |
---|
| 573 | ! stored in array "numpart_mpi(np)". This array is sorted from low to |
---|
| 574 | ! high values, along with a corresponding process ID array "idx_arr(np)". |
---|
| 575 | ! If the relative difference between the highest and lowest "numpart_mpi" value |
---|
| 576 | ! is above a threshold, particles are transferred from process idx_arr(np-1) |
---|
| 577 | ! to process idx_arr(0). Repeat for processes idx_arr(np-i) and idx_arr(i) |
---|
| 578 | ! for all valid i. |
---|
| 579 | ! Note: If np is an odd number, the process with the median |
---|
| 580 | ! number of particles is left unchanged |
---|
| 581 | ! |
---|
| 582 | ! VARIABLES |
---|
| 583 | ! itime input, current time |
---|
| 584 | !*********************************************************************** |
---|
| 585 | use com_mod |
---|
| 586 | |
---|
| 587 | implicit none |
---|
| 588 | |
---|
| 589 | integer, intent(in) :: itime |
---|
| 590 | real :: pmin,z |
---|
| 591 | integer :: i,jj,nn, num_part=1,m,imin, num_trans |
---|
| 592 | logical :: first_iter |
---|
[0a94e13] | 593 | integer,allocatable,dimension(:) :: idx_arr |
---|
[861805a] | 594 | real,allocatable,dimension(:) :: sorted ! TODO: we don't really need this |
---|
| 595 | |
---|
| 596 | ! Exit if running with only 1 process |
---|
| 597 | !************************************************************************ |
---|
| 598 | if (mp_np.eq.1) return |
---|
| 599 | |
---|
| 600 | ! All processes exchange information on number of particles |
---|
| 601 | !**************************************************************************** |
---|
[0a94e13] | 602 | allocate( idx_arr(0:mp_partgroup_np-1), sorted(0:mp_partgroup_np-1)) |
---|
[861805a] | 603 | |
---|
[0a94e13] | 604 | call MPI_Allgather(numpart, 1, MPI_INTEGER, npart_per_process, & |
---|
[861805a] | 605 | & 1, MPI_INTEGER, mp_comm_used, mp_ierr) |
---|
| 606 | |
---|
| 607 | |
---|
| 608 | ! Sort from lowest to highest |
---|
| 609 | ! Initial guess: correct order |
---|
[0a94e13] | 610 | sorted(:) = npart_per_process(:) |
---|
[861805a] | 611 | do i=0, mp_partgroup_np-1 |
---|
| 612 | idx_arr(i) = i |
---|
| 613 | end do |
---|
| 614 | |
---|
[0a94e13] | 615 | ! Do not rebalance particles for ipout=3 |
---|
| 616 | if (ipout.eq.3) return |
---|
| 617 | |
---|
[861805a] | 618 | ! For each successive element in index array, see if a lower value exists |
---|
| 619 | do i=0, mp_partgroup_np-2 |
---|
| 620 | pmin=sorted(i) |
---|
| 621 | imin=idx_arr(i) |
---|
| 622 | m=i+1 |
---|
| 623 | do jj=m, mp_partgroup_np-1 |
---|
| 624 | if (pmin.le.sorted(jj)) cycle |
---|
| 625 | z=pmin |
---|
| 626 | pmin=sorted(jj) |
---|
| 627 | sorted(jj)=z |
---|
| 628 | |
---|
| 629 | nn=imin |
---|
| 630 | imin=idx_arr(jj) |
---|
| 631 | idx_arr(jj)=nn |
---|
| 632 | end do |
---|
| 633 | sorted(i)=pmin |
---|
| 634 | idx_arr(i)=imin |
---|
| 635 | end do |
---|
| 636 | |
---|
| 637 | ! For each pair of processes, transfer particles if the difference is above a |
---|
| 638 | ! limit, and numpart at sending process large enough |
---|
| 639 | |
---|
| 640 | m=mp_partgroup_np-1 ! index for last sorted process (most particles) |
---|
| 641 | do i=0,mp_partgroup_np/2-1 |
---|
[0a94e13] | 642 | num_trans = npart_per_process(idx_arr(m)) - npart_per_process(idx_arr(i)) |
---|
[861805a] | 643 | if (mp_partid.eq.idx_arr(m).or.mp_partid.eq.idx_arr(i)) then |
---|
[0a94e13] | 644 | if ( npart_per_process(idx_arr(m)).gt.mp_min_redist.and.& |
---|
| 645 | & real(num_trans)/real(npart_per_process(idx_arr(m))).gt.mp_redist_fract) then |
---|
[16b61a5] | 646 | ! DBG |
---|
[0a94e13] | 647 | ! write(*,*) 'mp_partid, idx_arr(m), idx_arr(i), mp_min_redist, num_trans, npart_per_process', & |
---|
| 648 | ! &mp_partid, idx_arr(m), idx_arr(i), mp_min_redist, num_trans, npart_per_process |
---|
[16b61a5] | 649 | ! DBG |
---|
[861805a] | 650 | call mpif_redist_part(itime, idx_arr(m), idx_arr(i), num_trans/2) |
---|
| 651 | end if |
---|
| 652 | end if |
---|
| 653 | m=m-1 |
---|
| 654 | end do |
---|
| 655 | |
---|
[0a94e13] | 656 | deallocate(idx_arr, sorted) |
---|
[861805a] | 657 | |
---|
| 658 | end subroutine mpif_calculate_part_redist |
---|
| 659 | |
---|
| 660 | |
---|
| 661 | subroutine mpif_redist_part(itime, src_proc, dest_proc, num_trans) |
---|
| 662 | !*********************************************************************** |
---|
| 663 | ! Transfer particle properties between two arbitrary processes. |
---|
| 664 | ! |
---|
| 665 | ! VARIABLES |
---|
| 666 | ! |
---|
| 667 | ! itime input, current time |
---|
| 668 | ! src_proc input, ID of source process |
---|
| 669 | ! dest_proc input, ID of destination process |
---|
| 670 | ! num_trans input, number of particles to transfer |
---|
| 671 | ! |
---|
| 672 | !************************************************************************ |
---|
| 673 | use com_mod !TODO: ,only: nclass etc |
---|
| 674 | |
---|
| 675 | implicit none |
---|
| 676 | |
---|
| 677 | integer, intent(in) :: itime, src_proc, dest_proc, num_trans |
---|
| 678 | integer :: ll, ul ! lower and upper indices in arrays |
---|
| 679 | integer :: arr_sz ! size of temporary arrays |
---|
| 680 | integer :: mtag ! MPI message tag |
---|
| 681 | integer :: i, j, minpart, ipart, maxnumpart |
---|
| 682 | |
---|
[16b61a5] | 683 | ! Check for invalid input arguments |
---|
| 684 | !********************************** |
---|
| 685 | if (src_proc.eq.dest_proc) then |
---|
| 686 | write(*,*) '<mpi_mod::mpif_redist_part>: Error: & |
---|
| 687 | &src_proc.eq.dest_proc' |
---|
| 688 | stop |
---|
| 689 | end if |
---|
| 690 | |
---|
[861805a] | 691 | ! Measure time for MPI communications |
---|
| 692 | !************************************ |
---|
| 693 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 694 | |
---|
| 695 | ! Send particles |
---|
| 696 | !*************** |
---|
| 697 | if (mp_partid.eq.src_proc) then |
---|
| 698 | mtag = 2000 |
---|
| 699 | |
---|
| 700 | ! Array indices for the transferred particles |
---|
| 701 | ll=numpart-num_trans+1 |
---|
| 702 | ul=numpart |
---|
| 703 | |
---|
[16b61a5] | 704 | if (mp_dev_mode) then |
---|
| 705 | write(*,FMT='(72("#"))') |
---|
| 706 | write(*,*) "Sending ", num_trans, "particles (from/to)", src_proc, dest_proc |
---|
| 707 | write(*,*) "numpart before: ", numpart |
---|
| 708 | end if |
---|
[861805a] | 709 | |
---|
| 710 | call MPI_SEND(nclass(ll:ul),num_trans,& |
---|
| 711 | & MPI_INTEGER,dest_proc,mtag+1,mp_comm_used,mp_ierr) |
---|
| 712 | |
---|
| 713 | call MPI_SEND(npoint(ll:ul),num_trans,& |
---|
| 714 | & MPI_INTEGER,dest_proc,mtag+2,mp_comm_used,mp_ierr) |
---|
| 715 | |
---|
| 716 | call MPI_SEND(itra1(ll:ul),num_trans, & |
---|
| 717 | & MPI_INTEGER,dest_proc,mtag+3,mp_comm_used,mp_ierr) |
---|
| 718 | |
---|
| 719 | call MPI_SEND(idt(ll:ul),num_trans, & |
---|
| 720 | & MPI_INTEGER,dest_proc,mtag+4,mp_comm_used,mp_ierr) |
---|
| 721 | |
---|
| 722 | call MPI_SEND(itramem(ll:ul),num_trans, & |
---|
| 723 | & MPI_INTEGER,dest_proc,mtag+5,mp_comm_used,mp_ierr) |
---|
| 724 | |
---|
| 725 | call MPI_SEND(itrasplit(ll:ul),num_trans,& |
---|
| 726 | & MPI_INTEGER,dest_proc,mtag+6,mp_comm_used,mp_ierr) |
---|
| 727 | |
---|
| 728 | call MPI_SEND(xtra1(ll:ul),num_trans, & |
---|
| 729 | & mp_dp,dest_proc,mtag+7,mp_comm_used,mp_ierr) |
---|
| 730 | |
---|
| 731 | call MPI_SEND(ytra1(ll:ul),num_trans,& |
---|
| 732 | & mp_dp,dest_proc,mtag+8,mp_comm_used,mp_ierr) |
---|
| 733 | |
---|
| 734 | call MPI_SEND(ztra1(ll:ul),num_trans,& |
---|
| 735 | & mp_sp,dest_proc,mtag+9,mp_comm_used,mp_ierr) |
---|
| 736 | |
---|
| 737 | do j=1,nspec |
---|
| 738 | call MPI_SEND(xmass1(ll:ul,j),num_trans,& |
---|
| 739 | & mp_sp,dest_proc,mtag+(9+j),mp_comm_used,mp_ierr) |
---|
| 740 | end do |
---|
| 741 | |
---|
| 742 | ! Terminate transferred particles, update numpart |
---|
| 743 | itra1(ll:ul) = -999999999 |
---|
| 744 | |
---|
| 745 | numpart = numpart-num_trans |
---|
| 746 | |
---|
[16b61a5] | 747 | if (mp_dev_mode) then |
---|
| 748 | write(*,*) "numpart after: ", numpart |
---|
| 749 | write(*,FMT='(72("#"))') |
---|
| 750 | end if |
---|
[861805a] | 751 | |
---|
| 752 | else if (mp_partid.eq.dest_proc) then |
---|
| 753 | |
---|
| 754 | ! Receive particles |
---|
| 755 | !****************** |
---|
| 756 | mtag = 2000 |
---|
| 757 | ! Array indices for the transferred particles |
---|
| 758 | ll=numpart+1 |
---|
| 759 | ul=numpart+num_trans |
---|
| 760 | |
---|
[16b61a5] | 761 | if (mp_dev_mode) then |
---|
| 762 | write(*,FMT='(72("#"))') |
---|
| 763 | write(*,*) "Receiving ", num_trans, "particles (from/to)", src_proc, dest_proc |
---|
| 764 | write(*,*) "numpart before: ", numpart |
---|
| 765 | end if |
---|
[861805a] | 766 | |
---|
| 767 | ! We could receive the data directly at nclass(ll:ul) etc., but this leaves |
---|
| 768 | ! vacant spaces at lower indices. Using temporary arrays instead. |
---|
| 769 | arr_sz = ul-ll+1 |
---|
| 770 | allocate(itra1_tmp(arr_sz),npoint_tmp(arr_sz),nclass_tmp(arr_sz),& |
---|
| 771 | & idt_tmp(arr_sz),itramem_tmp(arr_sz),itrasplit_tmp(arr_sz),& |
---|
| 772 | & xtra1_tmp(arr_sz),ytra1_tmp(arr_sz),ztra1_tmp(arr_sz),& |
---|
| 773 | & xmass1_tmp(arr_sz, maxspec)) |
---|
| 774 | |
---|
| 775 | call MPI_RECV(nclass_tmp,num_trans,& |
---|
| 776 | & MPI_INTEGER,src_proc,mtag+1,mp_comm_used,mp_status,mp_ierr) |
---|
| 777 | |
---|
| 778 | call MPI_RECV(npoint_tmp,num_trans,& |
---|
| 779 | & MPI_INTEGER,src_proc,mtag+2,mp_comm_used,mp_status,mp_ierr) |
---|
| 780 | |
---|
| 781 | call MPI_RECV(itra1_tmp,num_trans, & |
---|
| 782 | & MPI_INTEGER,src_proc,mtag+3,mp_comm_used,mp_status,mp_ierr) |
---|
| 783 | |
---|
| 784 | call MPI_RECV(idt_tmp,num_trans, & |
---|
| 785 | & MPI_INTEGER,src_proc,mtag+4,mp_comm_used,mp_status,mp_ierr) |
---|
| 786 | |
---|
| 787 | call MPI_RECV(itramem_tmp,num_trans, & |
---|
| 788 | & MPI_INTEGER,src_proc,mtag+5,mp_comm_used,mp_status,mp_ierr) |
---|
| 789 | |
---|
| 790 | call MPI_RECV(itrasplit_tmp,num_trans,& |
---|
| 791 | & MPI_INTEGER,src_proc,mtag+6,mp_comm_used,mp_status,mp_ierr) |
---|
| 792 | |
---|
| 793 | call MPI_RECV(xtra1_tmp,num_trans, & |
---|
| 794 | & mp_dp,src_proc,mtag+7,mp_comm_used,mp_status,mp_ierr) |
---|
| 795 | |
---|
| 796 | call MPI_RECV(ytra1_tmp,num_trans,& |
---|
| 797 | & mp_dp,src_proc,mtag+8,mp_comm_used,mp_status,mp_ierr) |
---|
| 798 | |
---|
| 799 | call MPI_RECV(ztra1_tmp,num_trans,& |
---|
| 800 | & mp_sp,src_proc,mtag+9,mp_comm_used,mp_status,mp_ierr) |
---|
| 801 | |
---|
| 802 | do j=1,nspec |
---|
| 803 | call MPI_RECV(xmass1_tmp(:,j),num_trans,& |
---|
| 804 | & mp_sp,src_proc,mtag+(9+j),mp_comm_used,mp_status,mp_ierr) |
---|
| 805 | end do |
---|
| 806 | |
---|
| 807 | ! This is the maximum value numpart can possibly have after the transfer |
---|
| 808 | maxnumpart=numpart+num_trans |
---|
| 809 | |
---|
| 810 | ! Search for vacant space and copy from temporary storage |
---|
| 811 | !******************************************************** |
---|
| 812 | minpart=1 |
---|
| 813 | do i=1, num_trans |
---|
| 814 | ! Take into acount that we may have transferred invalid particles |
---|
[16b61a5] | 815 | if (itra1_tmp(i).ne.itime) cycle |
---|
[861805a] | 816 | do ipart=minpart,maxnumpart |
---|
| 817 | if (itra1(ipart).ne.itime) then |
---|
[16b61a5] | 818 | itra1(ipart) = itra1_tmp(i) |
---|
| 819 | npoint(ipart) = npoint_tmp(i) |
---|
| 820 | nclass(ipart) = nclass_tmp(i) |
---|
| 821 | idt(ipart) = idt_tmp(i) |
---|
| 822 | itramem(ipart) = itramem_tmp(i) |
---|
| 823 | itrasplit(ipart) = itrasplit_tmp(i) |
---|
| 824 | xtra1(ipart) = xtra1_tmp(i) |
---|
| 825 | ytra1(ipart) = ytra1_tmp(i) |
---|
| 826 | ztra1(ipart) = ztra1_tmp(i) |
---|
| 827 | xmass1(ipart,:) = xmass1_tmp(i,:) |
---|
[861805a] | 828 | goto 200 ! Storage space has been found, stop searching |
---|
| 829 | end if |
---|
[16b61a5] | 830 | ! :TODO: add check for if too many particles requiried |
---|
[861805a] | 831 | end do |
---|
[16b61a5] | 832 | 200 minpart=ipart+1 |
---|
[861805a] | 833 | end do |
---|
[16b61a5] | 834 | ! Increase numpart, if necessary |
---|
| 835 | numpart=max(numpart,ipart) |
---|
[861805a] | 836 | |
---|
[4c64400] | 837 | deallocate(itra1_tmp,npoint_tmp,nclass_tmp,idt_tmp,itramem_tmp,& |
---|
| 838 | &itrasplit_tmp,xtra1_tmp,ytra1_tmp,ztra1_tmp,xmass1_tmp) |
---|
[861805a] | 839 | |
---|
[16b61a5] | 840 | if (mp_dev_mode) then |
---|
| 841 | write(*,*) "numpart after: ", numpart |
---|
| 842 | write(*,FMT='(72("#"))') |
---|
| 843 | end if |
---|
[861805a] | 844 | |
---|
| 845 | else |
---|
| 846 | ! This routine should only be called by the two participating processes |
---|
| 847 | write(*,*) "ERROR: wrong process has entered mpi_mod::mpif_redist_part" |
---|
| 848 | stop |
---|
| 849 | return |
---|
| 850 | end if |
---|
| 851 | |
---|
| 852 | ! Measure time for MPI communications |
---|
| 853 | !************************************ |
---|
| 854 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 855 | |
---|
| 856 | end subroutine mpif_redist_part |
---|
| 857 | |
---|
| 858 | |
---|
[8a65cb0] | 859 | subroutine mpif_tm_send_vars |
---|
| 860 | !*********************************************************************** |
---|
| 861 | ! Distribute particle variables from pid0 to all processes. |
---|
| 862 | ! Called from timemanager |
---|
[6b22af9] | 863 | ! *NOT IN USE* at the moment, but can be useful for debugging |
---|
[8a65cb0] | 864 | ! |
---|
| 865 | !*********************************************************************** |
---|
| 866 | use com_mod |
---|
| 867 | |
---|
| 868 | implicit none |
---|
| 869 | |
---|
| 870 | integer :: i |
---|
| 871 | |
---|
| 872 | ! Time for MPI communications |
---|
| 873 | !**************************** |
---|
| 874 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 875 | |
---|
| 876 | ! Distribute variables, send from pid 0 to other processes |
---|
| 877 | !********************************************************************** |
---|
| 878 | |
---|
| 879 | ! integers |
---|
| 880 | if (lroot) then |
---|
| 881 | call MPI_SCATTER(npoint,numpart_mpi,MPI_INTEGER,MPI_IN_PLACE,& |
---|
[861805a] | 882 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 883 | if (mp_ierr /= 0) goto 600 |
---|
| 884 | call MPI_SCATTER(idt,numpart_mpi,MPI_INTEGER,MPI_IN_PLACE,& |
---|
[861805a] | 885 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 886 | if (mp_ierr /= 0) goto 600 |
---|
| 887 | call MPI_SCATTER(itra1,numpart_mpi,MPI_INTEGER,MPI_IN_PLACE,& |
---|
[861805a] | 888 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 889 | if (mp_ierr /= 0) goto 600 |
---|
| 890 | call MPI_SCATTER(nclass,numpart_mpi,MPI_INTEGER,MPI_IN_PLACE,& |
---|
[861805a] | 891 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 892 | if (mp_ierr /= 0) goto 600 |
---|
| 893 | call MPI_SCATTER(itramem,numpart_mpi,MPI_INTEGER,MPI_IN_PLACE,& |
---|
[861805a] | 894 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 895 | if (mp_ierr /= 0) goto 600 |
---|
| 896 | |
---|
| 897 | ! int2 |
---|
| 898 | call MPI_SCATTER(cbt,numpart_mpi,MPI_INTEGER2,MPI_IN_PLACE,& |
---|
[861805a] | 899 | & numpart_mpi,MPI_INTEGER2,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 900 | if (mp_ierr /= 0) goto 600 |
---|
| 901 | |
---|
| 902 | ! real |
---|
[fdc0f03] | 903 | call MPI_SCATTER(uap,numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 904 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 905 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 906 | call MPI_SCATTER(ucp,numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 907 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 908 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 909 | call MPI_SCATTER(uzp,numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 910 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 911 | if (mp_ierr /= 0) goto 600 |
---|
| 912 | |
---|
[fdc0f03] | 913 | call MPI_SCATTER(us,numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 914 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 915 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 916 | call MPI_SCATTER(vs,numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 917 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 918 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 919 | call MPI_SCATTER(ws,numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 920 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 921 | if (mp_ierr /= 0) goto 600 |
---|
| 922 | |
---|
| 923 | call MPI_SCATTER(xtra1,numpart_mpi,mp_dp,MPI_IN_PLACE,& |
---|
[861805a] | 924 | & numpart_mpi,mp_dp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 925 | if (mp_ierr /= 0) goto 600 |
---|
| 926 | call MPI_SCATTER(ytra1,numpart_mpi,mp_dp,MPI_IN_PLACE,& |
---|
[861805a] | 927 | & numpart_mpi,mp_dp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 928 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 929 | call MPI_SCATTER(ztra1,numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 930 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 931 | if (mp_ierr /= 0) goto 600 |
---|
| 932 | |
---|
| 933 | do i=1,nspec |
---|
[fdc0f03] | 934 | call MPI_SCATTER(xmass1(:,i),numpart_mpi,mp_sp,MPI_IN_PLACE,& |
---|
[861805a] | 935 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 936 | if (mp_ierr /= 0) goto 600 |
---|
| 937 | end do |
---|
| 938 | |
---|
| 939 | else ! (mp_pid >= 1) |
---|
| 940 | ! integers |
---|
| 941 | call MPI_SCATTER(npoint,numpart_mpi,MPI_INTEGER,npoint,& |
---|
[861805a] | 942 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 943 | if (mp_ierr /= 0) goto 600 |
---|
| 944 | call MPI_SCATTER(idt,numpart_mpi,MPI_INTEGER,idt,& |
---|
[861805a] | 945 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 946 | if (mp_ierr /= 0) goto 600 |
---|
| 947 | call MPI_SCATTER(itra1,numpart_mpi,MPI_INTEGER,itra1,& |
---|
[861805a] | 948 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 949 | if (mp_ierr /= 0) goto 600 |
---|
| 950 | call MPI_SCATTER(nclass,numpart_mpi,MPI_INTEGER,nclass,& |
---|
[861805a] | 951 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 952 | if (mp_ierr /= 0) goto 600 |
---|
| 953 | call MPI_SCATTER(itramem,numpart_mpi,MPI_INTEGER,itramem,& |
---|
[861805a] | 954 | & numpart_mpi,MPI_INTEGER,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 955 | if (mp_ierr /= 0) goto 600 |
---|
| 956 | |
---|
| 957 | ! int2 |
---|
| 958 | call MPI_SCATTER(cbt,numpart_mpi,MPI_INTEGER2,cbt,& |
---|
[861805a] | 959 | & numpart_mpi,MPI_INTEGER2,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 960 | if (mp_ierr /= 0) goto 600 |
---|
| 961 | |
---|
| 962 | ! reals |
---|
[fdc0f03] | 963 | call MPI_SCATTER(uap,numpart_mpi,mp_sp,uap,& |
---|
[861805a] | 964 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 965 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 966 | call MPI_SCATTER(ucp,numpart_mpi,mp_sp,ucp,& |
---|
[861805a] | 967 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 968 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 969 | call MPI_SCATTER(uzp,numpart_mpi,mp_sp,uzp,& |
---|
[861805a] | 970 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 971 | if (mp_ierr /= 0) goto 600 |
---|
| 972 | |
---|
[fdc0f03] | 973 | call MPI_SCATTER(us,numpart_mpi,mp_sp,us,& |
---|
[861805a] | 974 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 975 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 976 | call MPI_SCATTER(vs,numpart_mpi,mp_sp,vs,& |
---|
[861805a] | 977 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 978 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 979 | call MPI_SCATTER(ws,numpart_mpi,mp_sp,ws,& |
---|
[861805a] | 980 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 981 | if (mp_ierr /= 0) goto 600 |
---|
| 982 | |
---|
| 983 | call MPI_SCATTER(xtra1,numpart_mpi,mp_dp,xtra1,& |
---|
[861805a] | 984 | & numpart_mpi,mp_dp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 985 | if (mp_ierr /= 0) goto 600 |
---|
| 986 | call MPI_SCATTER(ytra1,numpart_mpi,mp_dp,ytra1,& |
---|
[861805a] | 987 | & numpart_mpi,mp_dp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 988 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 989 | call MPI_SCATTER(ztra1,numpart_mpi,mp_sp,ztra1,& |
---|
[861805a] | 990 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 991 | if (mp_ierr /= 0) goto 600 |
---|
| 992 | |
---|
| 993 | do i=1,nspec |
---|
[fdc0f03] | 994 | call MPI_SCATTER(xmass1(:,i),numpart_mpi,mp_sp,xmass1(:,i),& |
---|
[861805a] | 995 | & numpart_mpi,mp_sp,id_root,mp_comm_used,mp_ierr) |
---|
[8a65cb0] | 996 | if (mp_ierr /= 0) goto 600 |
---|
| 997 | end do |
---|
| 998 | |
---|
| 999 | end if !(lroot) |
---|
| 1000 | |
---|
| 1001 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 1002 | |
---|
| 1003 | goto 601 |
---|
| 1004 | |
---|
| 1005 | 600 write(*,*) "mpi_mod> mp_ierr \= 0", mp_ierr |
---|
| 1006 | stop |
---|
| 1007 | 601 end subroutine mpif_tm_send_vars |
---|
| 1008 | |
---|
| 1009 | |
---|
| 1010 | subroutine mpif_tm_collect_vars |
---|
| 1011 | !******************************************************************************* |
---|
| 1012 | ! Collect particle data to PID 0 from all processes. * |
---|
| 1013 | ! * |
---|
| 1014 | ! This can be called from timemanager_mpi, before caclulating * |
---|
| 1015 | ! concentrations/writing output grids. * |
---|
| 1016 | ! * |
---|
[6b22af9] | 1017 | ! Currently *not in use*, as each process calculates concentrations * |
---|
| 1018 | ! separately, but can be useful for debugging * |
---|
[8a65cb0] | 1019 | ! * |
---|
| 1020 | ! To use this routine (together with mpif_tm_send_vars) to transfer data * |
---|
| 1021 | ! to the MPI root process (useful for debugging), insert code like this * |
---|
| 1022 | ! (in timemanager_mpi.f90) * |
---|
| 1023 | ! * |
---|
[6b22af9] | 1024 | ! if (lroot) tot_numpart=numpart ! root stores total numpart * |
---|
[8a65cb0] | 1025 | ! call mpif_tm_send_dims * |
---|
| 1026 | ! if (numpart>1) then * |
---|
| 1027 | ! call mpif_tm_send_vars * |
---|
| 1028 | ! end if * |
---|
| 1029 | ! * |
---|
| 1030 | ! To collect data from all processes to root process after a parallel * |
---|
| 1031 | ! region: * |
---|
| 1032 | ! * |
---|
| 1033 | ! if (numpart>0) then * |
---|
[6b22af9] | 1034 | ! if (lroot) numpart = tot_numpart * |
---|
[8a65cb0] | 1035 | ! call mpif_tm_collect_vars * |
---|
| 1036 | ! end if * |
---|
| 1037 | ! * |
---|
[6b22af9] | 1038 | ! Then a section that begins with "if (lroot) ..." will behave like * |
---|
| 1039 | ! serial flexpart * |
---|
[8a65cb0] | 1040 | ! * |
---|
| 1041 | !******************************************************************************* |
---|
| 1042 | use com_mod !, only: numpart, nspec, itra1, npoint, nclass |
---|
| 1043 | |
---|
| 1044 | implicit none |
---|
| 1045 | |
---|
| 1046 | integer :: i |
---|
| 1047 | |
---|
| 1048 | logical :: use_mp_vars = .false.! use mp_... type allocated vars |
---|
| 1049 | logical :: use_in_place = .true.! use MPI_IN_PLACE to collect vars. |
---|
| 1050 | |
---|
| 1051 | |
---|
| 1052 | ! Time for MPI communications |
---|
| 1053 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 1054 | |
---|
| 1055 | ! Distribute variables, send from pid 0 to other processes |
---|
| 1056 | !********************************************************************** |
---|
| 1057 | |
---|
| 1058 | if (.not. use_mp_vars.and..not.use_in_place) then |
---|
| 1059 | |
---|
| 1060 | ! Integers: |
---|
| 1061 | call MPI_GATHER(npoint, numpart_mpi, MPI_INTEGER, npoint, & |
---|
| 1062 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1063 | if (mp_ierr /= 0) goto 600 |
---|
| 1064 | call MPI_GATHER(idt, numpart_mpi, MPI_INTEGER, idt, & |
---|
| 1065 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1066 | if (mp_ierr /= 0) goto 600 |
---|
| 1067 | call MPI_GATHER(itra1, numpart_mpi, MPI_INTEGER, itra1, & |
---|
| 1068 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1069 | if (mp_ierr /= 0) goto 600 |
---|
| 1070 | call MPI_GATHER(nclass, numpart_mpi, MPI_INTEGER, nclass, & |
---|
| 1071 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1072 | if (mp_ierr /= 0) goto 600 |
---|
| 1073 | call MPI_GATHER(itramem, numpart_mpi, MPI_INTEGER, itramem, & |
---|
| 1074 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1075 | if (mp_ierr /= 0) goto 600 |
---|
| 1076 | |
---|
| 1077 | ! int2 |
---|
| 1078 | call MPI_GATHER(cbt, numpart_mpi, MPI_INTEGER2, cbt, & |
---|
| 1079 | &numpart_mpi, MPI_INTEGER2, id_root, mp_comm_used, mp_ierr) |
---|
| 1080 | if (mp_ierr /= 0) goto 600 |
---|
| 1081 | |
---|
| 1082 | ! Reals: |
---|
[fdc0f03] | 1083 | call MPI_GATHER(uap, numpart_mpi, mp_sp, uap, & |
---|
| 1084 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1085 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1086 | call MPI_GATHER(ucp, numpart_mpi, mp_sp, ucp, & |
---|
| 1087 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1088 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1089 | call MPI_GATHER(uzp, numpart_mpi, mp_sp, uzp, & |
---|
| 1090 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1091 | if (mp_ierr /= 0) goto 600 |
---|
| 1092 | |
---|
[fdc0f03] | 1093 | call MPI_GATHER(us, numpart_mpi, mp_sp, us, & |
---|
| 1094 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1095 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1096 | call MPI_GATHER(vs, numpart_mpi, mp_sp, vs, & |
---|
| 1097 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1098 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1099 | call MPI_GATHER(ws, numpart_mpi, mp_sp, ws, & |
---|
| 1100 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1101 | if (mp_ierr /= 0) goto 600 |
---|
| 1102 | |
---|
| 1103 | |
---|
| 1104 | call MPI_GATHER(xtra1, numpart_mpi, mp_dp, xtra1, & |
---|
| 1105 | &numpart_mpi, mp_dp, id_root, mp_comm_used, mp_ierr) |
---|
| 1106 | if (mp_ierr /= 0) goto 600 |
---|
| 1107 | call MPI_GATHER(ytra1, numpart_mpi, mp_dp, ytra1, & |
---|
| 1108 | &numpart_mpi, mp_dp, id_root, mp_comm_used, mp_ierr) |
---|
| 1109 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1110 | call MPI_GATHER(ztra1, numpart_mpi, mp_sp, ztra1, & |
---|
| 1111 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1112 | if (mp_ierr /= 0) goto 600 |
---|
| 1113 | |
---|
| 1114 | do i=1, nspec |
---|
[fdc0f03] | 1115 | call MPI_GATHER(xmass1(:,i), numpart_mpi, mp_sp, xmass1(:,i), & |
---|
| 1116 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1117 | if (mp_ierr /= 0) goto 600 |
---|
| 1118 | |
---|
| 1119 | end do |
---|
| 1120 | |
---|
| 1121 | ! Use variables npoint etc directly for communications |
---|
| 1122 | !*********************************************************************** |
---|
| 1123 | else if (use_in_place.and..not.use_mp_vars) then |
---|
| 1124 | if (lroot) then |
---|
| 1125 | ! Integers: |
---|
| 1126 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, MPI_INTEGER, npoint, & |
---|
| 1127 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1128 | if (mp_ierr /= 0) goto 600 |
---|
| 1129 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, MPI_INTEGER, idt, & |
---|
| 1130 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1131 | if (mp_ierr /= 0) goto 600 |
---|
| 1132 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, MPI_INTEGER, itra1, & |
---|
| 1133 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1134 | if (mp_ierr /= 0) goto 600 |
---|
| 1135 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, MPI_INTEGER, nclass, & |
---|
| 1136 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1137 | if (mp_ierr /= 0) goto 600 |
---|
| 1138 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, MPI_INTEGER, itramem, & |
---|
| 1139 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1140 | if (mp_ierr /= 0) goto 600 |
---|
| 1141 | |
---|
| 1142 | ! int2 |
---|
| 1143 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, MPI_INTEGER2, cbt, & |
---|
| 1144 | &numpart_mpi, MPI_INTEGER2, id_root, mp_comm_used, mp_ierr) |
---|
| 1145 | if (mp_ierr /= 0) goto 600 |
---|
| 1146 | |
---|
| 1147 | ! Reals: |
---|
[fdc0f03] | 1148 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, uap, & |
---|
| 1149 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1150 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1151 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, ucp, & |
---|
| 1152 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1153 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1154 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, uzp, & |
---|
| 1155 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1156 | if (mp_ierr /= 0) goto 600 |
---|
| 1157 | |
---|
[fdc0f03] | 1158 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, us, & |
---|
| 1159 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1160 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1161 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, vs, & |
---|
| 1162 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1163 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1164 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, ws, & |
---|
| 1165 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1166 | if (mp_ierr /= 0) goto 600 |
---|
| 1167 | |
---|
| 1168 | |
---|
| 1169 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_dp, xtra1, & |
---|
| 1170 | &numpart_mpi, mp_dp, id_root, mp_comm_used, mp_ierr) |
---|
| 1171 | if (mp_ierr /= 0) goto 600 |
---|
| 1172 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_dp, ytra1, & |
---|
| 1173 | &numpart_mpi, mp_dp, id_root, mp_comm_used, mp_ierr) |
---|
| 1174 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1175 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, ztra1, & |
---|
| 1176 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1177 | if (mp_ierr /= 0) goto 600 |
---|
| 1178 | |
---|
| 1179 | do i=1, nspec |
---|
[fdc0f03] | 1180 | call MPI_GATHER(MPI_IN_PLACE, numpart_mpi, mp_sp, xmass1(:,i), & |
---|
| 1181 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1182 | if (mp_ierr /= 0) goto 600 |
---|
| 1183 | end do |
---|
| 1184 | |
---|
| 1185 | else ! (for mp_pid >= 1) |
---|
| 1186 | |
---|
| 1187 | ! Integers: |
---|
| 1188 | call MPI_GATHER(npoint, numpart_mpi, MPI_INTEGER, npoint, & |
---|
| 1189 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1190 | if (mp_ierr /= 0) goto 600 |
---|
| 1191 | call MPI_GATHER(idt, numpart_mpi, MPI_INTEGER, idt, & |
---|
| 1192 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1193 | if (mp_ierr /= 0) goto 600 |
---|
| 1194 | call MPI_GATHER(itra1, numpart_mpi, MPI_INTEGER, itra1, & |
---|
| 1195 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1196 | if (mp_ierr /= 0) goto 600 |
---|
| 1197 | call MPI_GATHER(nclass, numpart_mpi, MPI_INTEGER, nclass, & |
---|
| 1198 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1199 | if (mp_ierr /= 0) goto 600 |
---|
| 1200 | call MPI_GATHER(itramem, numpart_mpi, MPI_INTEGER, itramem, & |
---|
| 1201 | &numpart_mpi, MPI_INTEGER, id_root, mp_comm_used, mp_ierr) |
---|
| 1202 | if (mp_ierr /= 0) goto 600 |
---|
| 1203 | |
---|
| 1204 | ! int2 |
---|
| 1205 | call MPI_GATHER(cbt, numpart_mpi, MPI_INTEGER2, cbt, & |
---|
| 1206 | &numpart_mpi, MPI_INTEGER2, id_root, mp_comm_used, mp_ierr) |
---|
| 1207 | if (mp_ierr /= 0) goto 600 |
---|
| 1208 | |
---|
| 1209 | ! Reals: |
---|
[fdc0f03] | 1210 | call MPI_GATHER(uap, numpart_mpi, mp_sp, uap, & |
---|
| 1211 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1212 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1213 | call MPI_GATHER(ucp, numpart_mpi, mp_sp, ucp, & |
---|
| 1214 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1215 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1216 | call MPI_GATHER(uzp, numpart_mpi, mp_sp, uzp, & |
---|
| 1217 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1218 | if (mp_ierr /= 0) goto 600 |
---|
| 1219 | |
---|
[fdc0f03] | 1220 | call MPI_GATHER(us, numpart_mpi, mp_sp, us, & |
---|
| 1221 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1222 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1223 | call MPI_GATHER(vs, numpart_mpi, mp_sp, vs, & |
---|
| 1224 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1225 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1226 | call MPI_GATHER(ws, numpart_mpi, mp_sp, ws, & |
---|
| 1227 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1228 | if (mp_ierr /= 0) goto 600 |
---|
| 1229 | |
---|
| 1230 | |
---|
| 1231 | call MPI_GATHER(xtra1, numpart_mpi, mp_dp, xtra1, & |
---|
| 1232 | &numpart_mpi, mp_dp, id_root, mp_comm_used, mp_ierr) |
---|
| 1233 | if (mp_ierr /= 0) goto 600 |
---|
| 1234 | call MPI_GATHER(ytra1, numpart_mpi, mp_dp, ytra1, & |
---|
| 1235 | &numpart_mpi, mp_dp, id_root, mp_comm_used, mp_ierr) |
---|
| 1236 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1237 | call MPI_GATHER(ztra1, numpart_mpi, mp_sp, ztra1, & |
---|
| 1238 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1239 | if (mp_ierr /= 0) goto 600 |
---|
| 1240 | |
---|
| 1241 | do i=1, nspec |
---|
[fdc0f03] | 1242 | call MPI_GATHER(xmass1(:,i), numpart_mpi, mp_sp, xmass1(:,i), & |
---|
| 1243 | &numpart_mpi, mp_sp, id_root, mp_comm_used, mp_ierr) |
---|
[8a65cb0] | 1244 | if (mp_ierr /= 0) goto 600 |
---|
| 1245 | end do |
---|
| 1246 | end if ! (lroot) |
---|
| 1247 | end if ! (use_in_place) |
---|
| 1248 | |
---|
| 1249 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 1250 | |
---|
| 1251 | goto 601 |
---|
| 1252 | |
---|
| 1253 | 600 write(*,*) "mpi_mod> mp_ierr \= 0", mp_ierr |
---|
| 1254 | stop |
---|
| 1255 | 601 end subroutine mpif_tm_collect_vars |
---|
| 1256 | |
---|
| 1257 | |
---|
| 1258 | subroutine mpif_gf_send_vars(memstat) |
---|
| 1259 | !******************************************************************************* |
---|
| 1260 | ! DESCRIPTION |
---|
[5f9d14a] | 1261 | ! Distribute 'getfield' variables from reader process |
---|
[8a65cb0] | 1262 | ! |
---|
| 1263 | ! Called from timemanager |
---|
| 1264 | ! |
---|
| 1265 | ! NOTE |
---|
| 1266 | ! This subroutine distributes windfields read from the reader process to |
---|
[3b80e98] | 1267 | ! all other processes. Usually one set of fields are transfered, but at |
---|
[8a65cb0] | 1268 | ! first timestep when there are no fields in memory, both are transfered. |
---|
| 1269 | ! MPI_Bcast is used, so implicitly all processes are synchronized at this |
---|
| 1270 | ! step |
---|
| 1271 | ! |
---|
| 1272 | ! |
---|
| 1273 | !******************************************************************************* |
---|
| 1274 | use com_mod |
---|
| 1275 | use par_mod,only: numwfmem |
---|
| 1276 | |
---|
| 1277 | implicit none |
---|
| 1278 | |
---|
| 1279 | integer, intent(in) :: memstat |
---|
| 1280 | |
---|
| 1281 | ! Common array sizes used for communications |
---|
| 1282 | integer, parameter :: d3_size1 = nxmax*nymax*nzmax |
---|
| 1283 | integer, parameter :: d3_size2 = nxmax*nymax*nuvzmax |
---|
| 1284 | integer, parameter :: d2_size1 = nxmax*nymax |
---|
| 1285 | integer, parameter :: d2_size2 = nxmax*nymax*maxspec |
---|
| 1286 | integer, parameter :: d2_size3 = nxmax*nymax |
---|
| 1287 | integer, parameter :: d1_size1 = maxwf |
---|
| 1288 | |
---|
| 1289 | integer :: d3s1,d3s2,d2s1,d2s2 |
---|
| 1290 | |
---|
| 1291 | ! li,ui lower,upper indices of fields to be transfered (1-3, ui>=li) |
---|
| 1292 | integer :: li,ui |
---|
| 1293 | |
---|
| 1294 | ! First time routine is called the unchangeable fields will be transfered |
---|
| 1295 | logical :: first_call=.true. |
---|
| 1296 | |
---|
| 1297 | !********************************************************************** |
---|
| 1298 | |
---|
| 1299 | ! Sizes of arrays transfered |
---|
| 1300 | d3s1=d3_size1 |
---|
| 1301 | d3s2=d3_size2 |
---|
| 1302 | d2s1=d2_size1 |
---|
| 1303 | d2s2=d2_size2 |
---|
| 1304 | |
---|
| 1305 | ! Decide which field(s) need to be transfered |
---|
| 1306 | if (memstat.ge.32) then ! distribute 2 fields, to 1st/2nd indices |
---|
| 1307 | li=1; ui=2 |
---|
| 1308 | d3s1=2*d3_size1 |
---|
| 1309 | d3s2=2*d3_size2 |
---|
| 1310 | d2s1=2*d2_size1 |
---|
| 1311 | d2s2=2*d2_size2 |
---|
| 1312 | else if (memstat.eq.17) then ! distribute 1 field, place on 1st index |
---|
| 1313 | li=1; ui=1 |
---|
| 1314 | else if (memstat.eq.18) then ! distribute 1 field, place on 2nd index |
---|
| 1315 | li=2; ui=2 |
---|
| 1316 | else if (memstat.eq.19) then ! distribute 1 field, place on 3nd index |
---|
| 1317 | li=3; ui=3 |
---|
| 1318 | else |
---|
| 1319 | write(*,*) '#### mpi_mod::mpif_gf_send_vars> ERROR: ', & |
---|
| 1320 | & 'wrong value of memstat, exiting ####', memstat |
---|
| 1321 | stop |
---|
| 1322 | end if |
---|
| 1323 | |
---|
| 1324 | |
---|
| 1325 | ! Time for MPI communications |
---|
| 1326 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 1327 | |
---|
| 1328 | ! Send variables from getfield process (id_read) to other processes |
---|
| 1329 | !********************************************************************** |
---|
| 1330 | |
---|
[3b80e98] | 1331 | ! The non-reader processes need to know if cloud water was read. |
---|
[5f9d14a] | 1332 | call MPI_Bcast(readclouds,1,MPI_LOGICAL,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
| 1333 | if (mp_ierr /= 0) goto 600 |
---|
| 1334 | |
---|
[8a65cb0] | 1335 | ! Static fields/variables sent only at startup |
---|
| 1336 | if (first_call) then |
---|
[fdc0f03] | 1337 | call MPI_Bcast(oro(:,:),d2_size3,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1338 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1339 | call MPI_Bcast(excessoro(:,:),d2_size3,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1340 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1341 | call MPI_Bcast(lsm(:,:),d2_size3,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1342 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1343 | call MPI_Bcast(xlanduse(:,:,:),d2_size3*numclass,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1344 | if (mp_ierr /= 0) goto 600 |
---|
| 1345 | call MPI_Bcast(wftime,d1_size1,MPI_INTEGER,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
| 1346 | if (mp_ierr /= 0) goto 600 |
---|
| 1347 | call MPI_Bcast(numbwf,1,MPI_INTEGER,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
| 1348 | if (mp_ierr /= 0) goto 600 |
---|
| 1349 | call MPI_Bcast(nmixz,1,MPI_INTEGER,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
| 1350 | if (mp_ierr /= 0) goto 600 |
---|
| 1351 | call MPI_Bcast(height,nzmax,MPI_INTEGER,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[5f9d14a] | 1352 | if (mp_ierr /= 0) goto 600 |
---|
[8a65cb0] | 1353 | |
---|
| 1354 | first_call=.false. |
---|
| 1355 | endif |
---|
| 1356 | |
---|
[fdc0f03] | 1357 | call MPI_Bcast(uu(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1358 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1359 | call MPI_Bcast(vv(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1360 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1361 | call MPI_Bcast(uupol(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1362 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1363 | call MPI_Bcast(vvpol(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1364 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1365 | call MPI_Bcast(ww(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1366 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1367 | call MPI_Bcast(tt(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1368 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1369 | call MPI_Bcast(rho(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1370 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1371 | call MPI_Bcast(drhodz(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1372 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1373 | call MPI_Bcast(tth(:,:,:,li:ui),d3s2,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1374 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1375 | call MPI_Bcast(qvh(:,:,:,li:ui),d3s2,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1376 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1377 | call MPI_Bcast(qv(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1378 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1379 | call MPI_Bcast(pv(:,:,:,li:ui),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1380 | if (mp_ierr /= 0) goto 600 |
---|
| 1381 | call MPI_Bcast(clouds(:,:,:,li:ui),d3s1,MPI_INTEGER1,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[5f9d14a] | 1382 | if (mp_ierr /= 0) goto 600 |
---|
| 1383 | |
---|
| 1384 | ! cloud water/ice: |
---|
| 1385 | if (readclouds) then |
---|
[341f4b7] | 1386 | call MPI_Bcast(ctwc(:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[5f9d14a] | 1387 | if (mp_ierr /= 0) goto 600 |
---|
| 1388 | end if |
---|
[8a65cb0] | 1389 | |
---|
| 1390 | ! 2D fields |
---|
[fdc0f03] | 1391 | call MPI_Bcast(cloudsh(:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1392 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1393 | call MPI_Bcast(vdep(:,:,:,li:ui),d2s2,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1394 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1395 | call MPI_Bcast(ps(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1396 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1397 | call MPI_Bcast(sd(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1398 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1399 | call MPI_Bcast(tcc(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1400 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1401 | call MPI_Bcast(tt2(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1402 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1403 | call MPI_Bcast(td2(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1404 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1405 | call MPI_Bcast(lsprec(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1406 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1407 | call MPI_Bcast(convprec(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1408 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1409 | call MPI_Bcast(ustar(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1410 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1411 | call MPI_Bcast(wstar(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1412 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1413 | call MPI_Bcast(hmix(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1414 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1415 | call MPI_Bcast(tropopause(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1416 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1417 | call MPI_Bcast(oli(:,:,:,li:ui),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1418 | if (mp_ierr /= 0) goto 600 |
---|
| 1419 | |
---|
| 1420 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 1421 | |
---|
| 1422 | goto 601 |
---|
| 1423 | |
---|
| 1424 | 600 write(*,*) "mpi_mod> mp_ierr \= 0", mp_ierr |
---|
| 1425 | stop |
---|
| 1426 | |
---|
| 1427 | 601 end subroutine mpif_gf_send_vars |
---|
| 1428 | |
---|
| 1429 | |
---|
| 1430 | subroutine mpif_gf_send_vars_nest(memstat) |
---|
| 1431 | !*********************************************************************** |
---|
| 1432 | ! DESCRIPTION |
---|
| 1433 | ! Distribute 'getfield' variables from reader process to all processes. |
---|
| 1434 | ! For nested fields |
---|
| 1435 | ! |
---|
| 1436 | ! Called from timemanager |
---|
| 1437 | ! |
---|
| 1438 | ! NOTE |
---|
| 1439 | ! This subroutine distributes nested windfields from the reader process to |
---|
| 1440 | ! all other processes. Usually one set of fields is transfered, but at |
---|
| 1441 | ! first timestep when there are no fields in memory, both are transfered. |
---|
| 1442 | ! MPI_Bcast is used, so implicitly all processes are synchronized at this |
---|
| 1443 | ! step |
---|
| 1444 | ! |
---|
| 1445 | ! |
---|
| 1446 | !*********************************************************************** |
---|
| 1447 | use com_mod |
---|
[32b49c3] | 1448 | use par_mod,only: numwfmem |
---|
[8a65cb0] | 1449 | |
---|
| 1450 | implicit none |
---|
| 1451 | |
---|
| 1452 | integer, intent(in) :: memstat |
---|
| 1453 | integer :: i |
---|
| 1454 | ! li,ui lower,upper indices of fields to be transfered (1-3, ui>=li) |
---|
| 1455 | integer :: li,ui |
---|
| 1456 | |
---|
| 1457 | ! Common array sizes used for communications |
---|
[db712a8] | 1458 | integer :: d3_size1 = nxmaxn*nymaxn*nzmax |
---|
| 1459 | integer :: d3_size2 = nxmaxn*nymaxn*nuvzmax |
---|
| 1460 | integer :: d2_size1 = nxmaxn*nymaxn |
---|
| 1461 | integer :: d2_size2 = nxmaxn*nymaxn*maxspec |
---|
| 1462 | integer :: d2_size3 = nxmaxn*nymaxn |
---|
[8a65cb0] | 1463 | |
---|
| 1464 | integer :: d3s1,d3s2,d2s1,d2s2 |
---|
| 1465 | |
---|
| 1466 | ! First time routine is called the unchangeable fields will be transfered |
---|
| 1467 | logical :: first_call=.true. |
---|
| 1468 | |
---|
| 1469 | !********************************************************************** |
---|
| 1470 | |
---|
| 1471 | ! Sizes of arrays transfered |
---|
| 1472 | d3s1=d3_size1 |
---|
| 1473 | d3s2=d3_size2 |
---|
| 1474 | d2s1=d2_size1 |
---|
| 1475 | d2s2=d2_size2 |
---|
| 1476 | |
---|
| 1477 | ! Decide which field(s) need to be transfered |
---|
| 1478 | if (memstat.ge.32) then ! distribute 2 fields |
---|
| 1479 | li=1; ui=2 |
---|
| 1480 | d3s1=2*d3_size1 |
---|
| 1481 | d3s2=2*d3_size2 |
---|
| 1482 | d2s1=2*d2_size1 |
---|
| 1483 | d2s2=2*d2_size2 |
---|
| 1484 | else if (memstat.eq.17) then ! distribute 1 field, on 1st index |
---|
| 1485 | li=1; ui=1 |
---|
| 1486 | else if (memstat.eq.18) then ! distribute 1 field, on 2nd index |
---|
| 1487 | li=2; ui=2 |
---|
| 1488 | else if (memstat.eq.19) then ! distribute 1 field, on 3nd index |
---|
| 1489 | li=3; ui=3 |
---|
| 1490 | else |
---|
| 1491 | write(*,*) '#### mpi_mod::mpif_gf_send_vars_nest> ERROR: ', & |
---|
| 1492 | & 'wrong value of memstat, exiting ####', memstat |
---|
| 1493 | stop |
---|
| 1494 | end if |
---|
| 1495 | |
---|
| 1496 | |
---|
| 1497 | ! Time for MPI communications |
---|
| 1498 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 1499 | |
---|
| 1500 | ! Distribute variables, send from getfield process (id_read) to other |
---|
| 1501 | ! processes |
---|
| 1502 | !********************************************************************** |
---|
| 1503 | |
---|
[3b80e98] | 1504 | ! The non-reader processes need to know if cloud water was read. |
---|
[db712a8] | 1505 | call MPI_Bcast(readclouds_nest,maxnests,MPI_LOGICAL,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
| 1506 | if (mp_ierr /= 0) goto 600 |
---|
| 1507 | |
---|
[8a65cb0] | 1508 | ! Static fields/variables sent only at startup |
---|
| 1509 | if (first_call) then |
---|
[fdc0f03] | 1510 | call MPI_Bcast(oron(:,:,:),d2_size3,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1511 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1512 | call MPI_Bcast(excessoron(:,:,:),d2_size3,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1513 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1514 | call MPI_Bcast(lsmn(:,:,:),d2_size3,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1515 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1516 | call MPI_Bcast(xlandusen(:,:,:,:),d2_size3*numclass,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1517 | if (mp_ierr /= 0) goto 600 |
---|
| 1518 | first_call=.false. |
---|
| 1519 | end if |
---|
| 1520 | |
---|
| 1521 | ! MPI prefers contiguous arrays for sending (else a buffer is created), |
---|
[db712a8] | 1522 | ! hence the loop over nests |
---|
| 1523 | !********************************************************************** |
---|
[8a65cb0] | 1524 | do i=1, numbnests |
---|
| 1525 | ! 3D fields |
---|
[fdc0f03] | 1526 | call MPI_Bcast(uun(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1527 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1528 | call MPI_Bcast(vvn(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1529 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1530 | call MPI_Bcast(wwn(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1531 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1532 | call MPI_Bcast(ttn(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1533 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1534 | call MPI_Bcast(rhon(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1535 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1536 | call MPI_Bcast(drhodzn(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1537 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1538 | call MPI_Bcast(tthn(:,:,:,li:ui,i),d3s2,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1539 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1540 | call MPI_Bcast(qvhn(:,:,:,li:ui,i),d3s2,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1541 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1542 | call MPI_Bcast(qvn(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1543 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1544 | call MPI_Bcast(pvn(:,:,:,li:ui,i),d3s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1545 | if (mp_ierr /= 0) goto 600 |
---|
| 1546 | call MPI_Bcast(cloudsn(:,:,:,li:ui,i),d3s1,MPI_INTEGER1,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[db712a8] | 1547 | if (mp_ierr /= 0) goto 600 |
---|
| 1548 | |
---|
| 1549 | ! cloud water/ice: |
---|
[861805a] | 1550 | if (readclouds_nest(i)) then |
---|
| 1551 | ! call MPI_Bcast(icloud_stats(:,:,:,li:ui),d2s1*5,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
| 1552 | ! if (mp_ierr /= 0) goto 600 |
---|
| 1553 | call MPI_Bcast(ctwcn(:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
| 1554 | if (mp_ierr /= 0) goto 600 |
---|
| 1555 | end if |
---|
[8a65cb0] | 1556 | |
---|
| 1557 | ! 2D fields |
---|
[db712a8] | 1558 | call MPI_Bcast(cloudshn(:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1559 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1560 | call MPI_Bcast(vdepn(:,:,:,li:ui,i),d2s2,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1561 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1562 | call MPI_Bcast(psn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1563 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1564 | call MPI_Bcast(sdn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1565 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1566 | call MPI_Bcast(tccn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1567 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1568 | call MPI_Bcast(tt2n(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1569 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1570 | call MPI_Bcast(td2n(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1571 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1572 | call MPI_Bcast(lsprecn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1573 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1574 | call MPI_Bcast(convprecn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1575 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1576 | call MPI_Bcast(ustarn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1577 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1578 | call MPI_Bcast(wstarn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1579 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1580 | call MPI_Bcast(olin(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1581 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1582 | call MPI_Bcast(hmixn(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1583 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1584 | call MPI_Bcast(tropopausen(:,:,:,li:ui,i),d2s1,mp_sp,id_read,MPI_COMM_WORLD,mp_ierr) |
---|
[8a65cb0] | 1585 | if (mp_ierr /= 0) goto 600 |
---|
| 1586 | end do |
---|
| 1587 | |
---|
| 1588 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 1589 | |
---|
| 1590 | goto 601 |
---|
| 1591 | |
---|
| 1592 | 600 write(*,*) "mpi_mod> mp_ierr \= 0",mp_ierr |
---|
| 1593 | stop |
---|
| 1594 | |
---|
| 1595 | 601 end subroutine mpif_gf_send_vars_nest |
---|
| 1596 | |
---|
| 1597 | |
---|
| 1598 | subroutine mpif_gf_send_vars_async(memstat) |
---|
| 1599 | !******************************************************************************* |
---|
| 1600 | ! DESCRIPTION |
---|
| 1601 | ! Distribute 'getfield' variables from reader process to all processes. |
---|
[5f9d14a] | 1602 | ! Called from timemanager by reader process only |
---|
[8a65cb0] | 1603 | ! |
---|
| 1604 | ! NOTE |
---|
| 1605 | ! This version uses asynchronious sends. The newest fields are sent in the |
---|
| 1606 | ! background, so calculations can continue while |
---|
| 1607 | ! MPI communications are performed. |
---|
| 1608 | ! |
---|
| 1609 | ! The way the MPI tags/requests are sequenced means that this routine must |
---|
| 1610 | ! carefully match routine 'mpif_gf_recv_vars_async' |
---|
| 1611 | ! |
---|
| 1612 | ! VARIABLES |
---|
| 1613 | ! memstat -- input, for resolving pointer to windfield index being read |
---|
| 1614 | ! mind -- index where to place new fields |
---|
| 1615 | ! |
---|
| 1616 | ! |
---|
| 1617 | !******************************************************************************* |
---|
| 1618 | use com_mod |
---|
| 1619 | |
---|
| 1620 | implicit none |
---|
| 1621 | |
---|
| 1622 | integer, intent(in) :: memstat |
---|
| 1623 | integer :: mind |
---|
| 1624 | integer :: dest,i |
---|
| 1625 | |
---|
| 1626 | ! Common array sizes used for communications |
---|
| 1627 | integer :: d3s1 = nxmax*nymax*nzmax |
---|
| 1628 | integer :: d3s2 = nxmax*nymax*nuvzmax |
---|
| 1629 | integer :: d2s1 = nxmax*nymax |
---|
| 1630 | integer :: d2s2 = nxmax*nymax*maxspec |
---|
[32b49c3] | 1631 | ! integer :: d1s1 = maxwf |
---|
[8a65cb0] | 1632 | |
---|
| 1633 | !******************************************************************************* |
---|
| 1634 | |
---|
| 1635 | ! At the time the send is posted, the reader process is one step further |
---|
| 1636 | ! in the permutation of memstat compared with the receiving processes |
---|
| 1637 | |
---|
| 1638 | if (memstat.ge.32) then |
---|
| 1639 | ! last read was synchronous, to indices 1 and 2, 3 is free |
---|
| 1640 | write(*,*) "#### mpi_mod::mpif_gf_send_vars_async> ERROR: & |
---|
| 1641 | & memstat>=32 should never happen here." |
---|
| 1642 | stop |
---|
| 1643 | else if (memstat.eq.17) then |
---|
| 1644 | ! old fields on 1,2, send 3 |
---|
| 1645 | mind=3 |
---|
| 1646 | else if (memstat.eq.18) then |
---|
| 1647 | ! old fields on 2,3, send 1 |
---|
| 1648 | mind=1 |
---|
| 1649 | else if (memstat.eq.19) then |
---|
| 1650 | ! old fields on 3,1, send 2 |
---|
| 1651 | mind=2 |
---|
| 1652 | else |
---|
| 1653 | write(*,*) "#### mpi_mod::mpif_gf_send_vars_async> ERROR: & |
---|
| 1654 | & invalid memstat" |
---|
[aa8c34a] | 1655 | mind=-1 |
---|
| 1656 | stop |
---|
[8a65cb0] | 1657 | end if |
---|
| 1658 | |
---|
| 1659 | if (mp_dev_mode) then |
---|
| 1660 | if (mp_pid.ne.id_read) then |
---|
| 1661 | write(*,*) 'MPI_DEV_MODE: error calling mpif_gf_send_vars_async' |
---|
| 1662 | end if |
---|
| 1663 | end if |
---|
| 1664 | |
---|
| 1665 | if (mp_dev_mode) write(*,*) '## in mpif_gf_send_vars_async, memstat:', memstat |
---|
| 1666 | |
---|
| 1667 | ! Time for MPI communications |
---|
| 1668 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 1669 | |
---|
| 1670 | ! Loop over receiving processes, initiate data sending |
---|
| 1671 | !***************************************************** |
---|
| 1672 | |
---|
| 1673 | do dest=0,mp_np-1 ! mp_np-2 will also work if last proc reserved for reading |
---|
[78e62dc] | 1674 | ! TODO: use mp_partgroup_np here |
---|
[8a65cb0] | 1675 | if (dest.eq.id_read) cycle |
---|
| 1676 | i=dest*nvar_async |
---|
[fdc0f03] | 1677 | call MPI_Isend(uu(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1678 | if (mp_ierr /= 0) goto 600 |
---|
| 1679 | i=i+1 |
---|
[fdc0f03] | 1680 | call MPI_Isend(vv(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1681 | if (mp_ierr /= 0) goto 600 |
---|
| 1682 | i=i+1 |
---|
[fdc0f03] | 1683 | call MPI_Isend(uupol(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1684 | if (mp_ierr /= 0) goto 600 |
---|
| 1685 | i=i+1 |
---|
[fdc0f03] | 1686 | call MPI_Isend(vvpol(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1687 | if (mp_ierr /= 0) goto 600 |
---|
| 1688 | i=i+1 |
---|
[fdc0f03] | 1689 | call MPI_Isend(ww(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1690 | if (mp_ierr /= 0) goto 600 |
---|
| 1691 | i=i+1 |
---|
[fdc0f03] | 1692 | call MPI_Isend(tt(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1693 | if (mp_ierr /= 0) goto 600 |
---|
| 1694 | i=i+1 |
---|
[fdc0f03] | 1695 | call MPI_Isend(rho(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1696 | if (mp_ierr /= 0) goto 600 |
---|
| 1697 | i=i+1 |
---|
[fdc0f03] | 1698 | call MPI_Isend(drhodz(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1699 | if (mp_ierr /= 0) goto 600 |
---|
| 1700 | i=i+1 |
---|
[fdc0f03] | 1701 | call MPI_Isend(tth(:,:,:,mind),d3s2,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1702 | if (mp_ierr /= 0) goto 600 |
---|
| 1703 | i=i+1 |
---|
[fdc0f03] | 1704 | call MPI_Isend(qvh(:,:,:,mind),d3s2,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1705 | if (mp_ierr /= 0) goto 600 |
---|
| 1706 | i=i+1 |
---|
[fdc0f03] | 1707 | call MPI_Isend(qv(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1708 | if (mp_ierr /= 0) goto 600 |
---|
| 1709 | i=i+1 |
---|
[fdc0f03] | 1710 | call MPI_Isend(pv(:,:,:,mind),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1711 | if (mp_ierr /= 0) goto 600 |
---|
| 1712 | i=i+1 |
---|
| 1713 | call MPI_Isend(clouds(:,:,:,mind),d3s1,MPI_INTEGER1,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 1714 | i=i+1 |
---|
| 1715 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1716 | call MPI_Isend(cloudsh(:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1717 | if (mp_ierr /= 0) goto 600 |
---|
| 1718 | i=i+1 |
---|
[fdc0f03] | 1719 | call MPI_Isend(vdep(:,:,:,mind),d2s2,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1720 | if (mp_ierr /= 0) goto 600 |
---|
| 1721 | i=i+1 |
---|
[3b80e98] | 1722 | ! 15 |
---|
[fdc0f03] | 1723 | call MPI_Isend(ps(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1724 | if (mp_ierr /= 0) goto 600 |
---|
| 1725 | i=i+1 |
---|
[fdc0f03] | 1726 | call MPI_Isend(sd(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1727 | if (mp_ierr /= 0) goto 600 |
---|
| 1728 | i=i+1 |
---|
[fdc0f03] | 1729 | call MPI_Isend(tcc(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1730 | if (mp_ierr /= 0) goto 600 |
---|
| 1731 | i=i+1 |
---|
[fdc0f03] | 1732 | call MPI_Isend(tt2(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1733 | if (mp_ierr /= 0) goto 600 |
---|
| 1734 | i=i+1 |
---|
[fdc0f03] | 1735 | call MPI_Isend(td2(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1736 | if (mp_ierr /= 0) goto 600 |
---|
| 1737 | i=i+1 |
---|
[fdc0f03] | 1738 | call MPI_Isend(lsprec(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1739 | if (mp_ierr /= 0) goto 600 |
---|
| 1740 | i=i+1 |
---|
[fdc0f03] | 1741 | call MPI_Isend(convprec(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1742 | if (mp_ierr /= 0) goto 600 |
---|
| 1743 | i=i+1 |
---|
[fdc0f03] | 1744 | call MPI_Isend(ustar(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1745 | if (mp_ierr /= 0) goto 600 |
---|
| 1746 | i=i+1 |
---|
[fdc0f03] | 1747 | call MPI_Isend(wstar(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1748 | if (mp_ierr /= 0) goto 600 |
---|
| 1749 | i=i+1 |
---|
[fdc0f03] | 1750 | call MPI_Isend(hmix(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1751 | if (mp_ierr /= 0) goto 600 |
---|
| 1752 | i=i+1 |
---|
[3b80e98] | 1753 | ! 25 |
---|
[fdc0f03] | 1754 | call MPI_Isend(tropopause(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1755 | if (mp_ierr /= 0) goto 600 |
---|
| 1756 | i=i+1 |
---|
[fdc0f03] | 1757 | call MPI_Isend(oli(:,:,:,mind),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
[8a65cb0] | 1758 | if (mp_ierr /= 0) goto 600 |
---|
| 1759 | |
---|
[5f9d14a] | 1760 | ! Send cloud water if it exists. Increment counter always (as on receiving end) |
---|
| 1761 | if (readclouds) then |
---|
| 1762 | i=i+1 |
---|
[341f4b7] | 1763 | call MPI_Isend(ctwc(:,:,mind),d2s1,mp_sp,dest,tm1,& |
---|
[5f9d14a] | 1764 | &MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 1765 | if (mp_ierr /= 0) goto 600 |
---|
[d6a0977] | 1766 | end if |
---|
[8a65cb0] | 1767 | end do |
---|
| 1768 | |
---|
| 1769 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 1770 | |
---|
| 1771 | goto 601 |
---|
| 1772 | |
---|
[5f9d14a] | 1773 | 600 write(*,*) "#### mpi_mod::mpif_gf_send_vars_async> mp_ierr \= 0", mp_ierr |
---|
[8a65cb0] | 1774 | stop |
---|
| 1775 | |
---|
| 1776 | 601 end subroutine mpif_gf_send_vars_async |
---|
| 1777 | |
---|
| 1778 | |
---|
| 1779 | subroutine mpif_gf_recv_vars_async(memstat) |
---|
| 1780 | !******************************************************************************* |
---|
| 1781 | ! DESCRIPTION |
---|
[5f9d14a] | 1782 | ! Receive 'getfield' variables from reader process. |
---|
| 1783 | ! Called from timemanager by all processes except reader |
---|
[8a65cb0] | 1784 | ! |
---|
| 1785 | ! NOTE |
---|
| 1786 | ! This version uses asynchronious communications. |
---|
| 1787 | ! |
---|
| 1788 | ! VARIABLES |
---|
| 1789 | ! memstat -- input, used to resolve windfield index being received |
---|
| 1790 | ! |
---|
| 1791 | ! |
---|
| 1792 | !******************************************************************************* |
---|
| 1793 | use com_mod |
---|
| 1794 | |
---|
| 1795 | implicit none |
---|
| 1796 | |
---|
| 1797 | integer, intent(in) :: memstat |
---|
| 1798 | integer :: mind,j |
---|
| 1799 | |
---|
| 1800 | ! Common array sizes used for communications |
---|
| 1801 | integer :: d3s1 = nxmax*nymax*nzmax |
---|
| 1802 | integer :: d3s2 = nxmax*nymax*nuvzmax |
---|
| 1803 | integer :: d2s1 = nxmax*nymax |
---|
| 1804 | integer :: d2s2 = nxmax*nymax*maxspec |
---|
[861805a] | 1805 | !integer :: d1_size1 = maxwf |
---|
[8a65cb0] | 1806 | |
---|
| 1807 | ! integer :: d3s1,d3s2,d2s1,d2s2 |
---|
| 1808 | !******************************************************************************* |
---|
| 1809 | |
---|
| 1810 | ! At the time this immediate receive is posted, memstat is the state of |
---|
| 1811 | ! windfield indices at the previous/current time. From this, the future |
---|
| 1812 | ! state is deduced. |
---|
| 1813 | if (memstat.eq.32) then |
---|
| 1814 | ! last read was synchronous to indices 1 and 2, 3 is free |
---|
| 1815 | mind=3 |
---|
| 1816 | else if (memstat.eq.17) then |
---|
| 1817 | ! last read was asynchronous to index 3, 1 is free |
---|
| 1818 | mind=1 |
---|
| 1819 | else if (memstat.eq.18) then |
---|
| 1820 | ! last read was asynchronous to index 1, 2 is free |
---|
| 1821 | mind=2 |
---|
| 1822 | else if (memstat.eq.19) then |
---|
| 1823 | ! last read was asynchronous to index 2, 3 is free |
---|
| 1824 | mind=3 |
---|
[32b49c3] | 1825 | else |
---|
| 1826 | ! illegal state |
---|
| 1827 | write(*,*) 'mpi_mod> FLEXPART ERROR: Illegal memstat value. Exiting.' |
---|
| 1828 | stop |
---|
[8a65cb0] | 1829 | end if |
---|
| 1830 | |
---|
| 1831 | if (mp_dev_mode) then |
---|
| 1832 | if (mp_pid.eq.id_read) then |
---|
| 1833 | write(*,*) 'MPI_DEV_MODE: error calling mpif_gf_recv_vars_async' |
---|
| 1834 | end if |
---|
| 1835 | end if |
---|
| 1836 | |
---|
| 1837 | ! Time for MPI communications |
---|
| 1838 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 1839 | |
---|
| 1840 | if (mp_dev_mode) write(*,*) '## in mpif_gf_send_vars_async, memstat:', memstat |
---|
| 1841 | |
---|
| 1842 | ! Initiate receiving of data |
---|
| 1843 | !*************************** |
---|
| 1844 | |
---|
| 1845 | ! Get MPI tags/requests for communications |
---|
| 1846 | j=mp_pid*nvar_async |
---|
[fdc0f03] | 1847 | call MPI_Irecv(uu(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1848 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1849 | if (mp_ierr /= 0) goto 600 |
---|
| 1850 | j=j+1 |
---|
[fdc0f03] | 1851 | call MPI_Irecv(vv(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1852 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1853 | if (mp_ierr /= 0) goto 600 |
---|
| 1854 | j=j+1 |
---|
[fdc0f03] | 1855 | call MPI_Irecv(uupol(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1856 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1857 | if (mp_ierr /= 0) goto 600 |
---|
| 1858 | j=j+1 |
---|
[fdc0f03] | 1859 | call MPI_Irecv(vvpol(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1860 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1861 | if (mp_ierr /= 0) goto 600 |
---|
| 1862 | j=j+1 |
---|
[fdc0f03] | 1863 | call MPI_Irecv(ww(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1864 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1865 | if (mp_ierr /= 0) goto 600 |
---|
| 1866 | j=j+1 |
---|
[fdc0f03] | 1867 | call MPI_Irecv(tt(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1868 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1869 | if (mp_ierr /= 0) goto 600 |
---|
| 1870 | j=j+1 |
---|
[fdc0f03] | 1871 | call MPI_Irecv(rho(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1872 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1873 | if (mp_ierr /= 0) goto 600 |
---|
| 1874 | j=j+1 |
---|
[fdc0f03] | 1875 | call MPI_Irecv(drhodz(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1876 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1877 | if (mp_ierr /= 0) goto 600 |
---|
| 1878 | j=j+1 |
---|
[fdc0f03] | 1879 | call MPI_Irecv(tth(:,:,:,mind),d3s2,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1880 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1881 | if (mp_ierr /= 0) goto 600 |
---|
| 1882 | j=j+1 |
---|
[fdc0f03] | 1883 | call MPI_Irecv(qvh(:,:,:,mind),d3s2,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1884 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1885 | if (mp_ierr /= 0) goto 600 |
---|
| 1886 | j=j+1 |
---|
[fdc0f03] | 1887 | call MPI_Irecv(qv(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1888 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1889 | if (mp_ierr /= 0) goto 600 |
---|
[5f9d14a] | 1890 | j=j+1 |
---|
[fdc0f03] | 1891 | call MPI_Irecv(pv(:,:,:,mind),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1892 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1893 | if (mp_ierr /= 0) goto 600 |
---|
| 1894 | j=j+1 |
---|
[5f9d14a] | 1895 | call MPI_Irecv(clouds(:,:,:,mind),d3s1,MPI_INTEGER1,id_read,MPI_ANY_TAG,& |
---|
| 1896 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1897 | if (mp_ierr /= 0) goto 600 |
---|
| 1898 | j=j+1 |
---|
[fdc0f03] | 1899 | call MPI_Irecv(cloudsh(:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1900 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1901 | if (mp_ierr /= 0) goto 600 |
---|
| 1902 | j=j+1 |
---|
[fdc0f03] | 1903 | call MPI_Irecv(vdep(:,:,:,mind),d2s2,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1904 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1905 | if (mp_ierr /= 0) goto 600 |
---|
| 1906 | j=j+1 |
---|
[fdc0f03] | 1907 | call MPI_Irecv(ps(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1908 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1909 | if (mp_ierr /= 0) goto 600 |
---|
| 1910 | j=j+1 |
---|
[fdc0f03] | 1911 | call MPI_Irecv(sd(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1912 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1913 | if (mp_ierr /= 0) goto 600 |
---|
| 1914 | j=j+1 |
---|
[fdc0f03] | 1915 | call MPI_Irecv(tcc(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1916 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1917 | if (mp_ierr /= 0) goto 600 |
---|
| 1918 | j=j+1 |
---|
[fdc0f03] | 1919 | call MPI_Irecv(tt2(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1920 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1921 | if (mp_ierr /= 0) goto 600 |
---|
| 1922 | j=j+1 |
---|
[fdc0f03] | 1923 | call MPI_Irecv(td2(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1924 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1925 | if (mp_ierr /= 0) goto 600 |
---|
| 1926 | j=j+1 |
---|
[fdc0f03] | 1927 | call MPI_Irecv(lsprec(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1928 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1929 | if (mp_ierr /= 0) goto 600 |
---|
| 1930 | j=j+1 |
---|
[fdc0f03] | 1931 | call MPI_Irecv(convprec(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1932 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1933 | if (mp_ierr /= 0) goto 600 |
---|
[fdc0f03] | 1934 | call MPI_Irecv(ustar(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1935 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1936 | if (mp_ierr /= 0) goto 600 |
---|
| 1937 | j=j+1 |
---|
[fdc0f03] | 1938 | call MPI_Irecv(wstar(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1939 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1940 | if (mp_ierr /= 0) goto 600 |
---|
| 1941 | j=j+1 |
---|
[fdc0f03] | 1942 | call MPI_Irecv(hmix(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1943 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1944 | if (mp_ierr /= 0) goto 600 |
---|
| 1945 | j=j+1 |
---|
[fdc0f03] | 1946 | call MPI_Irecv(tropopause(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1947 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1948 | if (mp_ierr /= 0) goto 600 |
---|
| 1949 | j=j+1 |
---|
[fdc0f03] | 1950 | call MPI_Irecv(oli(:,:,:,mind),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[5f9d14a] | 1951 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
[8a65cb0] | 1952 | if (mp_ierr /= 0) goto 600 |
---|
| 1953 | |
---|
[5f9d14a] | 1954 | ! Post request for clwc. These data are possibly not sent, request must then be cancelled |
---|
| 1955 | ! For now assume that data at all steps either have or do not have water |
---|
| 1956 | if (readclouds) then |
---|
| 1957 | j=j+1 |
---|
[328fdf9] | 1958 | call MPI_Irecv(ctwc(:,:,mind),d2s1*5,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[d6a0977] | 1959 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 1960 | if (mp_ierr /= 0) goto 600 |
---|
[5f9d14a] | 1961 | end if |
---|
| 1962 | |
---|
| 1963 | |
---|
[8a65cb0] | 1964 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 1965 | |
---|
| 1966 | goto 601 |
---|
| 1967 | |
---|
[5f9d14a] | 1968 | 600 write(*,*) "#### mpi_mod::mpif_gf_recv_vars_async> MPI ERROR ", mp_ierr |
---|
[8a65cb0] | 1969 | stop |
---|
| 1970 | |
---|
| 1971 | 601 end subroutine mpif_gf_recv_vars_async |
---|
| 1972 | |
---|
| 1973 | |
---|
[3b80e98] | 1974 | subroutine mpif_gf_send_vars_nest_async(memstat) |
---|
| 1975 | !******************************************************************************* |
---|
| 1976 | ! DESCRIPTION |
---|
| 1977 | ! Distribute 'getfield' variables from reader process to all processes. |
---|
| 1978 | ! Called from timemanager by reader process only. |
---|
| 1979 | ! Version for nested wind fields |
---|
| 1980 | ! |
---|
| 1981 | ! NOTE |
---|
| 1982 | ! This version uses asynchronious sends. The newest fields are sent in the |
---|
| 1983 | ! background, so calculations can continue while |
---|
| 1984 | ! MPI communications are performed. |
---|
| 1985 | ! |
---|
| 1986 | ! The way the MPI tags/requests are sequenced means that this routine must |
---|
| 1987 | ! carefully match routine 'mpif_gf_recv_vars_async' |
---|
| 1988 | ! |
---|
| 1989 | ! VARIABLES |
---|
| 1990 | ! memstat -- input, for resolving pointer to windfield index being read |
---|
| 1991 | ! mind -- index where to place new fields |
---|
| 1992 | ! |
---|
| 1993 | ! TODO |
---|
| 1994 | ! Some unused arrays are currently sent (uupoln,..) |
---|
| 1995 | !******************************************************************************* |
---|
| 1996 | use com_mod |
---|
| 1997 | |
---|
| 1998 | implicit none |
---|
| 1999 | |
---|
| 2000 | integer, intent(in) :: memstat |
---|
| 2001 | integer :: mind |
---|
| 2002 | integer :: dest,i,k |
---|
| 2003 | |
---|
| 2004 | ! Common array sizes used for communications |
---|
| 2005 | integer :: d3s1 = nxmaxn*nymaxn*nzmax |
---|
| 2006 | integer :: d3s2 = nxmaxn*nymaxn*nuvzmax |
---|
| 2007 | integer :: d2s1 = nxmaxn*nymaxn |
---|
| 2008 | integer :: d2s2 = nxmaxn*nymaxn*maxspec |
---|
| 2009 | |
---|
| 2010 | !******************************************************************************* |
---|
| 2011 | |
---|
| 2012 | ! At the time the send is posted, the reader process is one step further |
---|
| 2013 | ! in the permutation of memstat compared with the receiving processes |
---|
| 2014 | |
---|
| 2015 | if (memstat.ge.32) then |
---|
| 2016 | ! last read was synchronous, to indices 1 and 2, 3 is free |
---|
| 2017 | write(*,*) "#### mpi_mod::mpif_gf_send_vars_nest_async> ERROR: & |
---|
| 2018 | & memstat>=32 should never happen here." |
---|
| 2019 | stop |
---|
| 2020 | else if (memstat.eq.17) then |
---|
| 2021 | ! old fields on 1,2, send 3 |
---|
| 2022 | mind=3 |
---|
| 2023 | else if (memstat.eq.18) then |
---|
| 2024 | ! old fields on 2,3, send 1 |
---|
| 2025 | mind=1 |
---|
| 2026 | else if (memstat.eq.19) then |
---|
| 2027 | ! old fields on 3,1, send 2 |
---|
| 2028 | mind=2 |
---|
| 2029 | else |
---|
| 2030 | write(*,*) "#### mpi_mod::mpif_gf_send_vars_nest_async> ERROR: & |
---|
| 2031 | & invalid memstat" |
---|
[aa8c34a] | 2032 | mind=-1 |
---|
| 2033 | stop |
---|
[3b80e98] | 2034 | end if |
---|
| 2035 | |
---|
| 2036 | if (mp_dev_mode) then |
---|
| 2037 | if (mp_pid.ne.id_read) then |
---|
| 2038 | write(*,*) 'MPI_DEV_MODE: error calling mpif_gf_send_vars_nest_async' |
---|
| 2039 | end if |
---|
| 2040 | end if |
---|
| 2041 | |
---|
| 2042 | if (mp_dev_mode) write(*,*) '## in mpif_gf_send_vars_nest_async, memstat:', memstat |
---|
| 2043 | |
---|
| 2044 | ! Time for MPI communications |
---|
| 2045 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 2046 | |
---|
| 2047 | ! Loop over receiving processes, initiate data sending |
---|
| 2048 | !***************************************************** |
---|
| 2049 | |
---|
| 2050 | do dest=0,mp_np-1 ! mp_np-2 will also work if last proc reserved for reading |
---|
| 2051 | ! TODO: use mp_partgroup_np here |
---|
| 2052 | if (dest.eq.id_read) cycle |
---|
| 2053 | do k=1, numbnests |
---|
[861805a] | 2054 | i=dest*nvar_async |
---|
| 2055 | call MPI_Isend(uun(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2056 | if (mp_ierr /= 0) goto 600 |
---|
| 2057 | i=i+1 |
---|
| 2058 | call MPI_Isend(vvn(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2059 | if (mp_ierr /= 0) goto 600 |
---|
| 2060 | i=i+1 |
---|
| 2061 | call MPI_Isend(wwn(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2062 | if (mp_ierr /= 0) goto 600 |
---|
| 2063 | i=i+1 |
---|
| 2064 | call MPI_Isend(ttn(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2065 | if (mp_ierr /= 0) goto 600 |
---|
| 2066 | i=i+1 |
---|
| 2067 | call MPI_Isend(rhon(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2068 | if (mp_ierr /= 0) goto 600 |
---|
| 2069 | i=i+1 |
---|
| 2070 | call MPI_Isend(drhodzn(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2071 | if (mp_ierr /= 0) goto 600 |
---|
| 2072 | i=i+1 |
---|
| 2073 | call MPI_Isend(tthn(:,:,:,mind,k),d3s2,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2074 | if (mp_ierr /= 0) goto 600 |
---|
| 2075 | i=i+1 |
---|
| 2076 | call MPI_Isend(qvhn(:,:,:,mind,k),d3s2,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2077 | if (mp_ierr /= 0) goto 600 |
---|
| 2078 | i=i+1 |
---|
| 2079 | call MPI_Isend(qvn(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2080 | if (mp_ierr /= 0) goto 600 |
---|
| 2081 | i=i+1 |
---|
| 2082 | call MPI_Isend(pvn(:,:,:,mind,k),d3s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2083 | if (mp_ierr /= 0) goto 600 |
---|
| 2084 | i=i+1 |
---|
| 2085 | call MPI_Isend(cloudsn(:,:,:,mind,k),d3s1,MPI_INTEGER1,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2086 | i=i+1 |
---|
| 2087 | if (mp_ierr /= 0) goto 600 |
---|
| 2088 | call MPI_Isend(cloudshn(:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2089 | if (mp_ierr /= 0) goto 600 |
---|
| 2090 | i=i+1 |
---|
| 2091 | call MPI_Isend(vdepn(:,:,:,mind,k),d2s2,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2092 | if (mp_ierr /= 0) goto 600 |
---|
| 2093 | i=i+1 |
---|
| 2094 | call MPI_Isend(psn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2095 | if (mp_ierr /= 0) goto 600 |
---|
| 2096 | i=i+1 |
---|
| 2097 | call MPI_Isend(sdn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2098 | if (mp_ierr /= 0) goto 600 |
---|
| 2099 | i=i+1 |
---|
[3b80e98] | 2100 | ! 15 |
---|
[861805a] | 2101 | call MPI_Isend(tccn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2102 | if (mp_ierr /= 0) goto 600 |
---|
| 2103 | i=i+1 |
---|
| 2104 | call MPI_Isend(tt2n(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2105 | if (mp_ierr /= 0) goto 600 |
---|
| 2106 | i=i+1 |
---|
| 2107 | call MPI_Isend(td2n(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2108 | if (mp_ierr /= 0) goto 600 |
---|
| 2109 | i=i+1 |
---|
| 2110 | call MPI_Isend(lsprecn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2111 | if (mp_ierr /= 0) goto 600 |
---|
| 2112 | i=i+1 |
---|
| 2113 | call MPI_Isend(convprecn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2114 | if (mp_ierr /= 0) goto 600 |
---|
| 2115 | i=i+1 |
---|
| 2116 | call MPI_Isend(ustarn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2117 | if (mp_ierr /= 0) goto 600 |
---|
| 2118 | i=i+1 |
---|
| 2119 | call MPI_Isend(wstarn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2120 | if (mp_ierr /= 0) goto 600 |
---|
| 2121 | i=i+1 |
---|
| 2122 | call MPI_Isend(hmixn(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2123 | if (mp_ierr /= 0) goto 600 |
---|
| 2124 | i=i+1 |
---|
| 2125 | call MPI_Isend(tropopausen(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2126 | if (mp_ierr /= 0) goto 600 |
---|
| 2127 | i=i+1 |
---|
| 2128 | call MPI_Isend(olin(:,:,:,mind,k),d2s1,mp_sp,dest,tm1,MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2129 | if (mp_ierr /= 0) goto 600 |
---|
[3b80e98] | 2130 | ! 25 |
---|
| 2131 | |
---|
| 2132 | ! Send cloud water if it exists. Increment counter always (as on receiving end) |
---|
[861805a] | 2133 | if (readclouds) then |
---|
| 2134 | i=i+1 |
---|
| 2135 | call MPI_Isend(ctwcn(:,:,mind,k),d2s1,mp_sp,dest,tm1,& |
---|
| 2136 | &MPI_COMM_WORLD,reqs(i),mp_ierr) |
---|
| 2137 | if (mp_ierr /= 0) goto 600 |
---|
| 2138 | end if |
---|
| 2139 | end do |
---|
[3b80e98] | 2140 | end do |
---|
| 2141 | |
---|
| 2142 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 2143 | |
---|
| 2144 | goto 601 |
---|
| 2145 | |
---|
| 2146 | 600 write(*,*) "#### mpi_mod::mpif_gf_send_vars_nest_async> mp_ierr \= 0", mp_ierr |
---|
| 2147 | stop |
---|
| 2148 | |
---|
| 2149 | 601 end subroutine mpif_gf_send_vars_nest_async |
---|
| 2150 | |
---|
| 2151 | |
---|
| 2152 | subroutine mpif_gf_recv_vars_nest_async(memstat) |
---|
| 2153 | !******************************************************************************* |
---|
| 2154 | ! DESCRIPTION |
---|
| 2155 | ! Receive 'getfield' variables from reader process. |
---|
| 2156 | ! Called from timemanager by all processes except reader |
---|
| 2157 | ! Version for nested wind fields |
---|
| 2158 | ! |
---|
| 2159 | ! NOTE |
---|
| 2160 | ! This version uses asynchronious communications. |
---|
| 2161 | ! |
---|
| 2162 | ! VARIABLES |
---|
| 2163 | ! memstat -- input, used to resolve windfield index being received |
---|
| 2164 | ! |
---|
| 2165 | ! |
---|
| 2166 | !******************************************************************************* |
---|
| 2167 | use com_mod |
---|
| 2168 | |
---|
| 2169 | implicit none |
---|
| 2170 | |
---|
| 2171 | integer, intent(in) :: memstat |
---|
| 2172 | integer :: mind,j,k |
---|
| 2173 | |
---|
| 2174 | ! Common array sizes used for communications |
---|
| 2175 | integer :: d3s1 = nxmaxn*nymaxn*nzmax |
---|
| 2176 | integer :: d3s2 = nxmaxn*nymaxn*nuvzmax |
---|
| 2177 | integer :: d2s1 = nxmaxn*nymaxn |
---|
| 2178 | integer :: d2s2 = nxmaxn*nymaxn*maxspec |
---|
| 2179 | |
---|
| 2180 | !******************************************************************************* |
---|
| 2181 | |
---|
| 2182 | ! At the time this immediate receive is posted, memstat is the state of |
---|
| 2183 | ! windfield indices at the previous/current time. From this, the future |
---|
| 2184 | ! state is deduced. |
---|
| 2185 | if (memstat.eq.32) then |
---|
| 2186 | ! last read was synchronous to indices 1 and 2, 3 is free |
---|
| 2187 | mind=3 |
---|
| 2188 | else if (memstat.eq.17) then |
---|
| 2189 | ! last read was asynchronous to index 3, 1 is free |
---|
| 2190 | mind=1 |
---|
| 2191 | else if (memstat.eq.18) then |
---|
| 2192 | ! last read was asynchronous to index 1, 2 is free |
---|
| 2193 | mind=2 |
---|
| 2194 | else if (memstat.eq.19) then |
---|
| 2195 | ! last read was asynchronous to index 2, 3 is free |
---|
| 2196 | mind=3 |
---|
| 2197 | else |
---|
| 2198 | ! illegal state |
---|
| 2199 | write(*,*) 'mpi_mod> FLEXPART ERROR: Illegal memstat value. Exiting.' |
---|
| 2200 | stop |
---|
| 2201 | end if |
---|
| 2202 | |
---|
| 2203 | if (mp_dev_mode) then |
---|
| 2204 | if (mp_pid.eq.id_read) then |
---|
| 2205 | write(*,*) 'MPI_DEV_MODE: error calling mpif_gf_recv_vars_async' |
---|
| 2206 | end if |
---|
| 2207 | end if |
---|
| 2208 | |
---|
| 2209 | ! Time for MPI communications |
---|
| 2210 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 2211 | |
---|
| 2212 | if (mp_dev_mode) write(*,*) '## in mpif_gf_send_vars_async, memstat:', memstat |
---|
| 2213 | |
---|
| 2214 | ! Initiate receiving of data |
---|
| 2215 | !*************************** |
---|
| 2216 | |
---|
| 2217 | do k=1, numbnests |
---|
| 2218 | ! Get MPI tags/requests for communications |
---|
[861805a] | 2219 | j=mp_pid*nvar_async |
---|
| 2220 | call MPI_Irecv(uun(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2221 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2222 | if (mp_ierr /= 0) goto 600 |
---|
[3b80e98] | 2223 | j=j+1 |
---|
[861805a] | 2224 | call MPI_Irecv(vvn(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2225 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2226 | if (mp_ierr /= 0) goto 600 |
---|
| 2227 | j=j+1 |
---|
| 2228 | call MPI_Irecv(wwn(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[3b80e98] | 2229 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2230 | if (mp_ierr /= 0) goto 600 |
---|
[861805a] | 2231 | j=j+1 |
---|
| 2232 | call MPI_Irecv(ttn(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2233 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2234 | if (mp_ierr /= 0) goto 600 |
---|
| 2235 | j=j+1 |
---|
| 2236 | call MPI_Irecv(rhon(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2237 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2238 | if (mp_ierr /= 0) goto 600 |
---|
| 2239 | j=j+1 |
---|
| 2240 | call MPI_Irecv(drhodzn(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2241 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2242 | if (mp_ierr /= 0) goto 600 |
---|
| 2243 | j=j+1 |
---|
| 2244 | call MPI_Irecv(tthn(:,:,:,mind,k),d3s2,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2245 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2246 | if (mp_ierr /= 0) goto 600 |
---|
| 2247 | j=j+1 |
---|
| 2248 | call MPI_Irecv(qvhn(:,:,:,mind,k),d3s2,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2249 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2250 | if (mp_ierr /= 0) goto 600 |
---|
| 2251 | j=j+1 |
---|
| 2252 | call MPI_Irecv(qvn(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2253 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2254 | if (mp_ierr /= 0) goto 600 |
---|
| 2255 | j=j+1 |
---|
| 2256 | call MPI_Irecv(pvn(:,:,:,mind,k),d3s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2257 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2258 | if (mp_ierr /= 0) goto 600 |
---|
| 2259 | j=j+1 |
---|
| 2260 | call MPI_Irecv(cloudsn(:,:,:,mind,k),d3s1,MPI_INTEGER1,id_read,MPI_ANY_TAG,& |
---|
| 2261 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2262 | if (mp_ierr /= 0) goto 600 |
---|
| 2263 | j=j+1 |
---|
| 2264 | call MPI_Irecv(cloudshn(:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2265 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2266 | if (mp_ierr /= 0) goto 600 |
---|
| 2267 | j=j+1 |
---|
| 2268 | call MPI_Irecv(vdepn(:,:,:,mind,k),d2s2,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2269 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2270 | if (mp_ierr /= 0) goto 600 |
---|
| 2271 | j=j+1 |
---|
| 2272 | call MPI_Irecv(psn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2273 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2274 | if (mp_ierr /= 0) goto 600 |
---|
| 2275 | j=j+1 |
---|
| 2276 | call MPI_Irecv(sdn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2277 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2278 | if (mp_ierr /= 0) goto 600 |
---|
| 2279 | j=j+1 |
---|
| 2280 | call MPI_Irecv(tccn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2281 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2282 | if (mp_ierr /= 0) goto 600 |
---|
| 2283 | j=j+1 |
---|
| 2284 | call MPI_Irecv(tt2n(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2285 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2286 | if (mp_ierr /= 0) goto 600 |
---|
| 2287 | j=j+1 |
---|
| 2288 | call MPI_Irecv(td2n(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2289 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2290 | if (mp_ierr /= 0) goto 600 |
---|
| 2291 | j=j+1 |
---|
| 2292 | call MPI_Irecv(lsprecn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2293 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2294 | if (mp_ierr /= 0) goto 600 |
---|
| 2295 | j=j+1 |
---|
| 2296 | call MPI_Irecv(convprecn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2297 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2298 | if (mp_ierr /= 0) goto 600 |
---|
| 2299 | call MPI_Irecv(ustarn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2300 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2301 | if (mp_ierr /= 0) goto 600 |
---|
| 2302 | j=j+1 |
---|
| 2303 | call MPI_Irecv(wstarn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2304 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2305 | if (mp_ierr /= 0) goto 600 |
---|
| 2306 | j=j+1 |
---|
| 2307 | call MPI_Irecv(hmixn(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2308 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2309 | if (mp_ierr /= 0) goto 600 |
---|
| 2310 | j=j+1 |
---|
| 2311 | call MPI_Irecv(tropopausen(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2312 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2313 | if (mp_ierr /= 0) goto 600 |
---|
| 2314 | j=j+1 |
---|
| 2315 | call MPI_Irecv(olin(:,:,:,mind,k),d2s1,mp_sp,id_read,MPI_ANY_TAG,& |
---|
| 2316 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2317 | if (mp_ierr /= 0) goto 600 |
---|
| 2318 | |
---|
| 2319 | ! Post request for clwc. These data are possibly not sent, request must then be cancelled |
---|
| 2320 | ! For now assume that data at all steps either have or do not have water |
---|
| 2321 | if (readclouds) then |
---|
| 2322 | j=j+1 |
---|
[328fdf9] | 2323 | call MPI_Irecv(ctwcn(:,:,mind,k),d2s1*5,mp_sp,id_read,MPI_ANY_TAG,& |
---|
[861805a] | 2324 | &MPI_COMM_WORLD,reqs(j),mp_ierr) |
---|
| 2325 | if (mp_ierr /= 0) goto 600 |
---|
| 2326 | end if |
---|
| 2327 | end do |
---|
[3b80e98] | 2328 | |
---|
| 2329 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 2330 | |
---|
| 2331 | goto 601 |
---|
| 2332 | |
---|
| 2333 | 600 write(*,*) "#### mpi_mod::mpif_gf_recv_vars_nest_async> MPI ERROR ", mp_ierr |
---|
| 2334 | stop |
---|
| 2335 | |
---|
| 2336 | 601 end subroutine mpif_gf_recv_vars_nest_async |
---|
| 2337 | |
---|
| 2338 | |
---|
[8a65cb0] | 2339 | subroutine mpif_gf_request |
---|
| 2340 | !******************************************************************************* |
---|
| 2341 | ! DESCRIPTION |
---|
| 2342 | ! Check for completion of MPI_Isend/MPI_Irecv data transfer. |
---|
| 2343 | ! |
---|
| 2344 | ! |
---|
| 2345 | ! NOTE |
---|
| 2346 | ! implicit synchronisation between all processes takes place here |
---|
| 2347 | ! |
---|
[3b80e98] | 2348 | ! TODO |
---|
[4c64400] | 2349 | ! NB: take into account nested wind fields by using a separate |
---|
| 2350 | ! subroutine mpif_gf_request_nest (and separate request arrays for the |
---|
| 2351 | ! nested variables) |
---|
[8a65cb0] | 2352 | ! |
---|
| 2353 | !******************************************************************************* |
---|
[32b49c3] | 2354 | ! use com_mod,only: readclouds |
---|
[5f9d14a] | 2355 | |
---|
[8a65cb0] | 2356 | implicit none |
---|
| 2357 | |
---|
[5f9d14a] | 2358 | |
---|
[32b49c3] | 2359 | integer :: n_req !,i |
---|
[5f9d14a] | 2360 | |
---|
| 2361 | !*********************************************************************** |
---|
| 2362 | |
---|
| 2363 | n_req=nvar_async*mp_np |
---|
| 2364 | |
---|
| 2365 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
[8a65cb0] | 2366 | |
---|
| 2367 | ! call MPI_Wait(rm1,MPI_STATUS_IGNORE,mp_ierr) |
---|
[5f9d14a] | 2368 | |
---|
| 2369 | ! TODO: cancel recv request if readclouds=.false. |
---|
| 2370 | ! if (readclouds) then |
---|
| 2371 | call MPI_Waitall(n_req,reqs,MPI_STATUSES_IGNORE,mp_ierr) |
---|
| 2372 | ! endif |
---|
[78e62dc] | 2373 | ! else |
---|
| 2374 | ! do i = 0, nvar_async*mp_np-1 |
---|
| 2375 | ! if (mod(i,27).eq.0 .or. mod(i,28).eq.0) then |
---|
| 2376 | ! call MPI_Cancel(reqs(i),mp_ierr) |
---|
| 2377 | ! cycle |
---|
| 2378 | ! end if |
---|
| 2379 | ! call MPI_Wait(reqs(i),MPI_STATUS_IGNORE,mp_ierr) |
---|
| 2380 | ! end do |
---|
| 2381 | ! end if |
---|
[5f9d14a] | 2382 | |
---|
[8a65cb0] | 2383 | if (mp_ierr /= 0) goto 600 |
---|
| 2384 | |
---|
[5f9d14a] | 2385 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
[8a65cb0] | 2386 | |
---|
| 2387 | goto 601 |
---|
| 2388 | |
---|
[5f9d14a] | 2389 | 600 write(*,*) "#### mpi_mod::mpif_gf_request> MPI ERROR ", mp_ierr |
---|
[8a65cb0] | 2390 | stop |
---|
| 2391 | |
---|
| 2392 | 601 end subroutine mpif_gf_request |
---|
| 2393 | |
---|
| 2394 | |
---|
| 2395 | subroutine mpif_tm_reduce_grid |
---|
| 2396 | !*********************************************************************** |
---|
| 2397 | ! Collect grid variable to PID 0, adding from all processes. |
---|
| 2398 | ! |
---|
| 2399 | ! NOTE |
---|
| 2400 | ! - Older MPI implementations may lack the MPI_IN_PLACE operation. |
---|
| 2401 | ! If so use 1) below and change unc_mod |
---|
| 2402 | ! |
---|
| 2403 | !*********************************************************************** |
---|
| 2404 | use com_mod |
---|
| 2405 | use unc_mod |
---|
| 2406 | use par_mod |
---|
| 2407 | |
---|
| 2408 | implicit none |
---|
| 2409 | |
---|
| 2410 | integer :: grid_size2d,grid_size3d |
---|
| 2411 | integer, parameter :: rcpt_size=maxreceptor*maxspec |
---|
| 2412 | |
---|
| 2413 | !********************************************************************** |
---|
| 2414 | grid_size2d=numxgrid*numygrid*maxspec* & |
---|
| 2415 | & maxpointspec_act*nclassunc*maxageclass |
---|
| 2416 | grid_size3d=numxgrid*numygrid*numzgrid*maxspec* & |
---|
| 2417 | & maxpointspec_act*nclassunc*maxageclass |
---|
| 2418 | |
---|
| 2419 | |
---|
| 2420 | ! Time for MPI communications |
---|
| 2421 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 2422 | |
---|
| 2423 | ! 1) Using a separate grid (gridunc0) for received values |
---|
[fdc0f03] | 2424 | ! call MPI_Reduce(gridunc, gridunc0, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2425 | ! & mp_comm_used, mp_ierr) |
---|
| 2426 | ! if (mp_ierr /= 0) goto 600 |
---|
| 2427 | |
---|
| 2428 | ! 2) Using in-place reduction |
---|
[6ecb30a] | 2429 | |
---|
| 2430 | !!!-------------------------------------------------------------------- |
---|
| 2431 | !!! DJM - 2017-05-09 change - MPI_IN_PLACE option for MPI_Reduce() causes |
---|
| 2432 | !!! severe numerical problems in some cases. I'm guessing there are memory |
---|
| 2433 | !!! overrun issues in this complex code, but have so far failed to identify |
---|
| 2434 | !!! a specific problem. And/or, when one searches the Internet for this |
---|
| 2435 | !!! issue, there is "some" hint that the implementation may be buggy. |
---|
| 2436 | !!! |
---|
| 2437 | !!! At this point, with the following CPP USE_MPIINPLACE directive, the |
---|
| 2438 | !!! default behaviour will be to not use the MPI_IN_PLACE option. |
---|
| 2439 | !!! Users will have to compile with -DUSE_MPIINPLACE if they want that option. |
---|
| 2440 | !!! Introduction of the CPP directives also requires that the code be compiled |
---|
| 2441 | !!! with the "-x f95-cpp-input" option. |
---|
| 2442 | !!! |
---|
| 2443 | !!! Modification of this section requires the addition of array gridunc0, which |
---|
| 2444 | !!! requires an allocation in outgrid_init.f90 and initial declaration in |
---|
| 2445 | !!! unc_mod.f90. |
---|
| 2446 | !!!-------------------------------------------------------------------- |
---|
| 2447 | |
---|
| 2448 | #ifdef USE_MPIINPLACE |
---|
| 2449 | |
---|
[8a65cb0] | 2450 | if (lroot) then |
---|
[fdc0f03] | 2451 | call MPI_Reduce(MPI_IN_PLACE, gridunc, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2452 | & mp_comm_used, mp_ierr) |
---|
| 2453 | if (mp_ierr /= 0) goto 600 |
---|
| 2454 | else |
---|
[79abee9] | 2455 | call MPI_Reduce(gridunc, 0, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2456 | & mp_comm_used, mp_ierr) |
---|
| 2457 | end if |
---|
| 2458 | |
---|
[328fdf9] | 2459 | ! Receptor concentrations |
---|
| 2460 | if (lroot) then |
---|
| 2461 | call MPI_Reduce(MPI_IN_PLACE,creceptor,rcpt_size,mp_sp,MPI_SUM,id_root, & |
---|
| 2462 | & mp_comm_used,mp_ierr) |
---|
| 2463 | if (mp_ierr /= 0) goto 600 |
---|
| 2464 | else |
---|
| 2465 | call MPI_Reduce(creceptor,0,rcpt_size,mp_sp,MPI_SUM,id_root, & |
---|
| 2466 | & mp_comm_used,mp_ierr) |
---|
| 2467 | end if |
---|
| 2468 | |
---|
[6ecb30a] | 2469 | #else |
---|
| 2470 | |
---|
| 2471 | call MPI_Reduce(gridunc, gridunc0, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
| 2472 | & mp_comm_used, mp_ierr) |
---|
[328fdf9] | 2473 | if (mp_ierr /= 0) goto 600 |
---|
[6ecb30a] | 2474 | if (lroot) gridunc = gridunc0 |
---|
| 2475 | |
---|
[328fdf9] | 2476 | call MPI_Reduce(creceptor, creceptor0,rcpt_size,mp_sp,MPI_SUM,id_root, & |
---|
| 2477 | & mp_comm_used,mp_ierr) |
---|
| 2478 | if (mp_ierr /= 0) goto 600 |
---|
| 2479 | if (lroot) creceptor = creceptor0 |
---|
| 2480 | |
---|
[6ecb30a] | 2481 | #endif |
---|
| 2482 | |
---|
[8a65cb0] | 2483 | if ((WETDEP).and.(ldirect.gt.0)) then |
---|
[6a678e3] | 2484 | call MPI_Reduce(wetgridunc, wetgridunc0, grid_size2d, mp_cp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2485 | & mp_comm_used, mp_ierr) |
---|
| 2486 | if (mp_ierr /= 0) goto 600 |
---|
| 2487 | end if |
---|
| 2488 | |
---|
| 2489 | if ((DRYDEP).and.(ldirect.gt.0)) then |
---|
[6a678e3] | 2490 | call MPI_Reduce(drygridunc, drygridunc0, grid_size2d, mp_cp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2491 | & mp_comm_used, mp_ierr) |
---|
| 2492 | if (mp_ierr /= 0) goto 600 |
---|
| 2493 | end if |
---|
| 2494 | |
---|
| 2495 | |
---|
| 2496 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 2497 | |
---|
| 2498 | goto 601 |
---|
| 2499 | |
---|
| 2500 | 600 write(*,*) "mpi_mod> mp_ierr \= 0", mp_ierr |
---|
| 2501 | stop |
---|
| 2502 | |
---|
| 2503 | 601 end subroutine mpif_tm_reduce_grid |
---|
| 2504 | |
---|
| 2505 | |
---|
| 2506 | subroutine mpif_tm_reduce_grid_nest |
---|
| 2507 | !*********************************************************************** |
---|
| 2508 | ! Collect nested grids to PID 0, adding from all processes. |
---|
| 2509 | ! |
---|
| 2510 | ! NOTE |
---|
| 2511 | ! - Compiling with 'fcheck=all' (gfortran) will cause a run-time error |
---|
| 2512 | ! as wetgriduncn0 is only allocated for root process. Possibly valid |
---|
| 2513 | ! MPI code but invalid gfortran. |
---|
| 2514 | ! |
---|
| 2515 | !*********************************************************************** |
---|
| 2516 | use com_mod |
---|
| 2517 | use unc_mod |
---|
| 2518 | use par_mod |
---|
| 2519 | |
---|
| 2520 | implicit none |
---|
| 2521 | |
---|
| 2522 | integer :: grid_size2d,grid_size3d |
---|
| 2523 | |
---|
| 2524 | !********************************************************************** |
---|
[5f9d14a] | 2525 | |
---|
[8a65cb0] | 2526 | grid_size3d=numxgridn*numygridn*numzgrid*maxspec* & |
---|
| 2527 | & maxpointspec_act*nclassunc*maxageclass |
---|
| 2528 | grid_size2d=numxgridn*numygridn*maxspec* & |
---|
| 2529 | & maxpointspec_act*nclassunc*maxageclass |
---|
| 2530 | |
---|
| 2531 | |
---|
| 2532 | ! Time for MPI communications |
---|
| 2533 | if (mp_measure_time) call mpif_mtime('commtime',0) |
---|
| 2534 | |
---|
| 2535 | ! Using a separate grid (gridunc0) for received values, for debugging |
---|
[fdc0f03] | 2536 | ! call MPI_Reduce(griduncn, griduncn0, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2537 | ! & mp_comm_used, mp_ierr) |
---|
| 2538 | ! if (mp_ierr /= 0) goto 600 |
---|
| 2539 | |
---|
[0ecc1fe] | 2540 | #ifdef USE_MPIINPLACE |
---|
[8a65cb0] | 2541 | ! Using in-place reduction |
---|
| 2542 | if (lroot) then |
---|
[fdc0f03] | 2543 | call MPI_Reduce(MPI_IN_PLACE, griduncn, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2544 | & mp_comm_used, mp_ierr) |
---|
| 2545 | if (mp_ierr /= 0) goto 600 |
---|
| 2546 | else |
---|
[79abee9] | 2547 | call MPI_Reduce(griduncn, 0, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2548 | & mp_comm_used, mp_ierr) |
---|
[0ecc1fe] | 2549 | if (mp_ierr /= 0) goto 600 |
---|
[8a65cb0] | 2550 | end if |
---|
| 2551 | |
---|
[0ecc1fe] | 2552 | #else |
---|
| 2553 | call MPI_Reduce(griduncn, griduncn0, grid_size3d, mp_sp, MPI_SUM, id_root, & |
---|
| 2554 | & mp_comm_used, mp_ierr) |
---|
| 2555 | if (mp_ierr /= 0) goto 600 |
---|
| 2556 | if (lroot) griduncn = griduncn0 |
---|
| 2557 | |
---|
| 2558 | #endif |
---|
| 2559 | |
---|
[8a65cb0] | 2560 | if ((WETDEP).and.(ldirect.gt.0)) then |
---|
[6a678e3] | 2561 | call MPI_Reduce(wetgriduncn, wetgriduncn0, grid_size2d, mp_cp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2562 | & mp_comm_used, mp_ierr) |
---|
| 2563 | if (mp_ierr /= 0) goto 600 |
---|
| 2564 | end if |
---|
| 2565 | |
---|
| 2566 | if ((DRYDEP).and.(ldirect.gt.0)) then |
---|
[6a678e3] | 2567 | call MPI_Reduce(drygriduncn, drygriduncn0, grid_size2d, mp_cp, MPI_SUM, id_root, & |
---|
[8a65cb0] | 2568 | & mp_comm_used, mp_ierr) |
---|
| 2569 | if (mp_ierr /= 0) goto 600 |
---|
| 2570 | end if |
---|
| 2571 | |
---|
| 2572 | if (mp_measure_time) call mpif_mtime('commtime',1) |
---|
| 2573 | |
---|
| 2574 | goto 601 |
---|
| 2575 | |
---|
| 2576 | 600 write(*,*) "mpi_mod> mp_ierr \= 0", mp_ierr |
---|
| 2577 | stop |
---|
| 2578 | |
---|
| 2579 | 601 end subroutine mpif_tm_reduce_grid_nest |
---|
| 2580 | |
---|
| 2581 | |
---|
| 2582 | subroutine mpif_mtime(ident,imode) |
---|
| 2583 | !*********************************************************************** |
---|
| 2584 | ! Measure CPU/Wall time in various parts of the code |
---|
| 2585 | ! |
---|
| 2586 | ! VARIABLES |
---|
| 2587 | ! ident character, identifies routine to measure |
---|
| 2588 | ! imode integer, 0:start clock(s) 1: stop clock(s) |
---|
| 2589 | ! |
---|
| 2590 | !*********************************************************************** |
---|
| 2591 | implicit none |
---|
| 2592 | |
---|
| 2593 | character(LEN=*), intent(in) :: ident |
---|
| 2594 | integer, intent(in) :: imode |
---|
| 2595 | |
---|
[5f9d14a] | 2596 | !*********************************************************************** |
---|
| 2597 | |
---|
[8a65cb0] | 2598 | select case(ident) |
---|
| 2599 | |
---|
| 2600 | case ('timemanager') |
---|
| 2601 | if (imode.eq.0) then |
---|
| 2602 | call cpu_time(tm_tot_beg) |
---|
| 2603 | else |
---|
| 2604 | call cpu_time(tm_tot_end) |
---|
| 2605 | tm_tot_total = tm_tot_total + (tm_tot_end-tm_tot_beg) |
---|
| 2606 | end if |
---|
| 2607 | |
---|
| 2608 | case ('wetdepo') |
---|
| 2609 | if (imode.eq.0) then |
---|
| 2610 | mp_wetdepo_wtime_beg = mpi_wtime() |
---|
| 2611 | call cpu_time(mp_wetdepo_time_beg) |
---|
| 2612 | else |
---|
| 2613 | mp_wetdepo_wtime_end = mpi_wtime() |
---|
| 2614 | call cpu_time(mp_wetdepo_time_end) |
---|
| 2615 | |
---|
| 2616 | mp_wetdepo_wtime_total = mp_wetdepo_wtime_total + & |
---|
| 2617 | &(mp_wetdepo_wtime_end - mp_wetdepo_wtime_beg) |
---|
| 2618 | mp_wetdepo_time_total = mp_wetdepo_time_total + & |
---|
| 2619 | &(mp_wetdepo_time_end - mp_wetdepo_time_beg) |
---|
| 2620 | end if |
---|
| 2621 | |
---|
[38b7917] | 2622 | case ('advance') |
---|
| 2623 | if (imode.eq.0) then |
---|
| 2624 | mp_advance_wtime_beg = mpi_wtime() |
---|
| 2625 | else |
---|
| 2626 | mp_advance_wtime_end = mpi_wtime() |
---|
| 2627 | |
---|
| 2628 | mp_advance_wtime_total = mp_advance_wtime_total + & |
---|
| 2629 | &(mp_advance_wtime_end - mp_advance_wtime_beg) |
---|
| 2630 | end if |
---|
| 2631 | |
---|
[8a65cb0] | 2632 | case ('getfields') |
---|
| 2633 | if (imode.eq.0) then |
---|
| 2634 | mp_getfields_wtime_beg = mpi_wtime() |
---|
| 2635 | call cpu_time(mp_getfields_time_beg) |
---|
| 2636 | else |
---|
| 2637 | mp_getfields_wtime_end = mpi_wtime() |
---|
| 2638 | call cpu_time(mp_getfields_time_end) |
---|
| 2639 | |
---|
| 2640 | mp_getfields_wtime_total = mp_getfields_wtime_total + & |
---|
| 2641 | &(mp_getfields_wtime_end - mp_getfields_wtime_beg) |
---|
| 2642 | mp_getfields_time_total = mp_getfields_time_total + & |
---|
| 2643 | &(mp_getfields_time_end - mp_getfields_time_beg) |
---|
| 2644 | end if |
---|
| 2645 | |
---|
| 2646 | case ('partloop1') |
---|
| 2647 | if (imode.eq.0) then |
---|
| 2648 | call cpu_time(tm_nploop_beg) |
---|
| 2649 | else |
---|
| 2650 | call cpu_time(tm_nploop_end) |
---|
| 2651 | tm_nploop_total = tm_nploop_total + (tm_nploop_end - tm_nploop_beg) |
---|
| 2652 | end if |
---|
| 2653 | |
---|
| 2654 | case ('conccalc') |
---|
| 2655 | if (imode.eq.0) then |
---|
| 2656 | call cpu_time(mp_conccalc_time_beg) |
---|
| 2657 | else |
---|
| 2658 | call cpu_time(mp_conccalc_time_end) |
---|
| 2659 | mp_conccalc_time_total = mp_conccalc_time_total + mp_conccalc_time_end - & |
---|
| 2660 | &mp_conccalc_time_beg |
---|
| 2661 | end if |
---|
[38b7917] | 2662 | |
---|
[8a65cb0] | 2663 | case ('rootonly') |
---|
| 2664 | if (imode.eq.0) then |
---|
| 2665 | call cpu_time(mp_root_time_beg) |
---|
| 2666 | mp_root_wtime_beg = mpi_wtime() |
---|
| 2667 | else |
---|
| 2668 | call cpu_time(mp_root_time_end) |
---|
| 2669 | mp_root_wtime_end = mpi_wtime() |
---|
| 2670 | mp_root_time_total = mp_root_time_total + & |
---|
| 2671 | &(mp_root_time_end - mp_root_time_beg) |
---|
| 2672 | mp_root_wtime_total = mp_root_wtime_total + & |
---|
| 2673 | &(mp_root_wtime_end - mp_root_wtime_beg) |
---|
| 2674 | end if |
---|
| 2675 | |
---|
| 2676 | case ('iotime') |
---|
| 2677 | if (imode.eq.0) then |
---|
| 2678 | mp_io_wtime_beg = mpi_wtime() |
---|
| 2679 | call cpu_time(mp_io_time_beg) |
---|
| 2680 | else |
---|
| 2681 | mp_io_wtime_end = mpi_wtime() |
---|
| 2682 | call cpu_time(mp_io_time_end) |
---|
| 2683 | |
---|
| 2684 | mp_io_wtime_total = mp_io_wtime_total + (mp_io_wtime_end - & |
---|
| 2685 | & mp_io_wtime_beg) |
---|
| 2686 | mp_io_time_total = mp_io_time_total + (mp_io_time_end - & |
---|
| 2687 | & mp_io_time_beg) |
---|
| 2688 | end if |
---|
| 2689 | |
---|
[78e62dc] | 2690 | case ('verttransform') |
---|
| 2691 | if (imode.eq.0) then |
---|
| 2692 | mp_vt_wtime_beg = mpi_wtime() |
---|
| 2693 | call cpu_time(mp_vt_time_beg) |
---|
| 2694 | else |
---|
| 2695 | mp_vt_wtime_end = mpi_wtime() |
---|
| 2696 | call cpu_time(mp_vt_time_end) |
---|
| 2697 | |
---|
| 2698 | mp_vt_wtime_total = mp_vt_wtime_total + (mp_vt_wtime_end - & |
---|
| 2699 | & mp_vt_wtime_beg) |
---|
| 2700 | mp_vt_time_total = mp_vt_time_total + (mp_vt_time_end - & |
---|
| 2701 | & mp_vt_time_beg) |
---|
| 2702 | end if |
---|
| 2703 | |
---|
[0a94e13] | 2704 | case ('readwind') |
---|
| 2705 | if (imode.eq.0) then |
---|
| 2706 | call cpu_time(mp_readwind_time_beg) |
---|
| 2707 | mp_readwind_wtime_beg = mpi_wtime() |
---|
| 2708 | else |
---|
| 2709 | call cpu_time(mp_readwind_time_end) |
---|
| 2710 | mp_readwind_wtime_end = mpi_wtime() |
---|
| 2711 | |
---|
| 2712 | mp_readwind_time_total = mp_readwind_time_total + & |
---|
| 2713 | &(mp_readwind_time_end - mp_readwind_time_beg) |
---|
| 2714 | mp_readwind_wtime_total = mp_readwind_wtime_total + & |
---|
| 2715 | &(mp_readwind_wtime_end - mp_readwind_wtime_beg) |
---|
| 2716 | end if |
---|
[8a65cb0] | 2717 | |
---|
| 2718 | case ('commtime') |
---|
| 2719 | if (imode.eq.0) then |
---|
| 2720 | call cpu_time(mp_comm_time_beg) |
---|
| 2721 | mp_comm_wtime_beg = mpi_wtime() |
---|
| 2722 | else |
---|
| 2723 | call cpu_time(mp_comm_time_end) |
---|
| 2724 | mp_comm_wtime_end = mpi_wtime() |
---|
| 2725 | mp_comm_time_total = mp_comm_time_total + & |
---|
| 2726 | &(mp_comm_time_end - mp_comm_time_beg) |
---|
| 2727 | mp_comm_wtime_total = mp_comm_wtime_total + & |
---|
| 2728 | &(mp_comm_wtime_end - mp_comm_wtime_beg) |
---|
| 2729 | end if |
---|
| 2730 | |
---|
| 2731 | case ('flexpart') |
---|
| 2732 | if (imode.eq.0) then |
---|
| 2733 | mp_total_wtime_beg=mpi_wtime() |
---|
| 2734 | else |
---|
| 2735 | mp_total_wtime_end=mpi_wtime() |
---|
| 2736 | mp_total_wtime_total = mp_total_wtime_end-mp_total_wtime_beg |
---|
| 2737 | end if |
---|
| 2738 | |
---|
| 2739 | case default |
---|
| 2740 | write(*,*) 'mpi_mod::mpif_mtime> WARNING: unknown case identifier' |
---|
| 2741 | |
---|
| 2742 | end select |
---|
| 2743 | |
---|
| 2744 | |
---|
| 2745 | end subroutine mpif_mtime |
---|
| 2746 | |
---|
| 2747 | |
---|
| 2748 | subroutine mpif_finalize |
---|
| 2749 | !*********************************************************************** |
---|
| 2750 | ! Finalize MPI |
---|
| 2751 | ! Optionally print detailed time diagnostics |
---|
| 2752 | !*********************************************************************** |
---|
| 2753 | implicit none |
---|
| 2754 | |
---|
[32b49c3] | 2755 | integer :: ip !,j,r |
---|
[8a65cb0] | 2756 | |
---|
[5f9d14a] | 2757 | !*********************************************************************** |
---|
| 2758 | |
---|
[78e62dc] | 2759 | IF (mp_measure_time) THEN |
---|
[8a65cb0] | 2760 | do ip=0, mp_np-1 |
---|
| 2761 | call MPI_BARRIER(MPI_COMM_WORLD, mp_ierr) |
---|
| 2762 | |
---|
| 2763 | if (ip == mp_pid) then |
---|
| 2764 | write(*,FMT='(72("#"))') |
---|
| 2765 | write(*,FMT='(A60,I3)') 'STATISTICS FOR MPI PROCESS:', mp_pid |
---|
| 2766 | if (ip == 0) then |
---|
| 2767 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR ROOT-ONLY (PID 0) & |
---|
| 2768 | &SECTIONS: ', mp_root_time_total |
---|
| 2769 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR ROOT-ONLY (PID 0) & |
---|
| 2770 | &SECTIONS:', mp_root_wtime_total |
---|
| 2771 | end if |
---|
| 2772 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR FLEXPART: ' & |
---|
| 2773 | &, mp_total_wtime_total |
---|
| 2774 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR MPI COMMUNICATIONS:',& |
---|
| 2775 | & mp_comm_time_total |
---|
| 2776 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR MPI COMMUNICATIONS:',& |
---|
| 2777 | & mp_comm_wtime_total |
---|
| 2778 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR MPI BARRIERS:',& |
---|
| 2779 | & mp_barrier_time_total |
---|
| 2780 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR MPI BARRIERS:',& |
---|
| 2781 | & mp_barrier_wtime_total |
---|
| 2782 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR TIMEMANAGER:',& |
---|
| 2783 | & tm_tot_total |
---|
| 2784 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR TIMEMANAGER/NUMPART LOOP:',& |
---|
| 2785 | & tm_nploop_total |
---|
| 2786 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR ADVANCE:',& |
---|
| 2787 | & mp_advance_wtime_total |
---|
| 2788 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR GETFIELDS:',& |
---|
| 2789 | & mp_getfields_wtime_total |
---|
| 2790 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR GETFIELDS:',& |
---|
| 2791 | & mp_getfields_time_total |
---|
[328fdf9] | 2792 | ! write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR READWIND:',& |
---|
| 2793 | ! & mp_readwind_wtime_total |
---|
| 2794 | ! write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR READWIND:',& |
---|
| 2795 | ! & mp_readwind_time_total |
---|
[8a65cb0] | 2796 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR FILE IO:',& |
---|
| 2797 | & mp_io_wtime_total |
---|
| 2798 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR FILE IO:',& |
---|
| 2799 | & mp_io_time_total |
---|
| 2800 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR WETDEPO:',& |
---|
| 2801 | & mp_wetdepo_wtime_total |
---|
| 2802 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR WETDEPO:',& |
---|
| 2803 | & mp_wetdepo_time_total |
---|
| 2804 | write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR CONCCALC:',& |
---|
| 2805 | & mp_conccalc_time_total |
---|
[861805a] | 2806 | ! write(*,FMT='(A60,TR1,F9.2)') 'TOTAL WALL TIME FOR VERTTRANSFORM:',& |
---|
| 2807 | ! & mp_vt_wtime_total |
---|
| 2808 | ! write(*,FMT='(A60,TR1,F9.2)') 'TOTAL CPU TIME FOR VERTTRANSFORM:',& |
---|
| 2809 | ! & mp_vt_time_total |
---|
[16b61a5] | 2810 | ! NB: the 'flush' function is possibly a gfortran-specific extension, |
---|
| 2811 | ! comment out if it gives problems |
---|
| 2812 | ! call flush() |
---|
[8a65cb0] | 2813 | end if |
---|
| 2814 | end do |
---|
| 2815 | end if |
---|
| 2816 | |
---|
[78e62dc] | 2817 | ! This call to barrier is for correctly formatting output |
---|
| 2818 | call MPI_BARRIER(MPI_COMM_WORLD, mp_ierr) |
---|
| 2819 | |
---|
[f55fdce] | 2820 | if (lroot.and.mp_measure_time) then |
---|
[78e62dc] | 2821 | write(*,FMT='(72("#"))') |
---|
| 2822 | WRITE(*,*) "To turn off output of time measurements, set " |
---|
| 2823 | WRITE(*,*) " mp_measure_time=.false." |
---|
| 2824 | WRITE(*,*) "in file mpi_mod.f90" |
---|
| 2825 | write(*,FMT='(72("#"))') |
---|
| 2826 | end if |
---|
| 2827 | |
---|
[8a65cb0] | 2828 | ! j=mp_pid*nvar_async |
---|
| 2829 | ! In the implementation with 3 fields, the processes may have posted |
---|
| 2830 | ! MPI_Irecv requests that should be cancelled here |
---|
[78e62dc] | 2831 | ! if (.not.lmp_sync) then |
---|
| 2832 | ! r=mp_pid*nvar_async |
---|
| 2833 | ! do j=r,r+nvar_async-1 |
---|
| 2834 | ! call MPI_Cancel(j,mp_ierr) |
---|
| 2835 | ! if (mp_ierr /= 0) write(*,*) '#### mpif_finalize::MPI_Cancel> ERROR ####' |
---|
| 2836 | ! end do |
---|
| 2837 | ! end if |
---|
[8a65cb0] | 2838 | |
---|
| 2839 | call MPI_FINALIZE(mp_ierr) |
---|
| 2840 | if (mp_ierr /= 0) then |
---|
[5f9d14a] | 2841 | write(*,*) '#### mpif_finalize::MPI_FINALIZE> MPI ERROR ', mp_ierr, ' ####' |
---|
[8a65cb0] | 2842 | stop |
---|
| 2843 | end if |
---|
| 2844 | |
---|
| 2845 | |
---|
[861805a] | 2846 | end subroutine mpif_finalize |
---|
[8a65cb0] | 2847 | |
---|
| 2848 | |
---|
[861805a] | 2849 | subroutine get_lun(my_lun) |
---|
[8a65cb0] | 2850 | !*********************************************************************** |
---|
| 2851 | ! get_lun: |
---|
| 2852 | ! Starting from 100, get next free logical unit number |
---|
| 2853 | !*********************************************************************** |
---|
| 2854 | |
---|
[861805a] | 2855 | implicit none |
---|
[8a65cb0] | 2856 | |
---|
[861805a] | 2857 | integer, intent(inout) :: my_lun |
---|
| 2858 | integer, save :: free_lun=100 |
---|
| 2859 | logical :: exists, iopen |
---|
[8a65cb0] | 2860 | |
---|
[5f9d14a] | 2861 | !*********************************************************************** |
---|
| 2862 | |
---|
[861805a] | 2863 | loop1: do |
---|
| 2864 | inquire(UNIT=free_lun, EXIST=exists, OPENED=iopen) |
---|
| 2865 | if (exists .and. .not.iopen) exit loop1 |
---|
| 2866 | free_lun = free_lun+1 |
---|
| 2867 | end do loop1 |
---|
| 2868 | my_lun = free_lun |
---|
[8a65cb0] | 2869 | |
---|
[861805a] | 2870 | end subroutine get_lun |
---|
[8a65cb0] | 2871 | |
---|
| 2872 | |
---|
[861805a] | 2873 | subroutine write_data_dbg(array_in, array_name, tstep, ident) |
---|
[8a65cb0] | 2874 | !*********************************************************************** |
---|
[78e62dc] | 2875 | ! Write one-dimensional arrays to file (for debugging purposes) |
---|
[8a65cb0] | 2876 | !*********************************************************************** |
---|
[861805a] | 2877 | implicit none |
---|
[8a65cb0] | 2878 | |
---|
[861805a] | 2879 | real, intent(in), dimension(:) :: array_in |
---|
| 2880 | integer, intent(in) :: tstep |
---|
| 2881 | integer :: lios |
---|
| 2882 | character(LEN=*), intent(in) :: ident, array_name |
---|
[8a65cb0] | 2883 | |
---|
[861805a] | 2884 | character(LEN=8) :: c_ts |
---|
| 2885 | character(LEN=40) :: fn_1, fn_2 |
---|
[8a65cb0] | 2886 | |
---|
[5f9d14a] | 2887 | !*********************************************************************** |
---|
| 2888 | |
---|
[861805a] | 2889 | write(c_ts, FMT='(I8.8,BZ)') tstep |
---|
| 2890 | fn_1='-'//trim(adjustl(c_ts))//'-'//trim(ident) |
---|
| 2891 | write(c_ts, FMT='(I2.2,BZ)') mp_np |
---|
| 2892 | fn_2= trim(adjustl(array_name))//trim(adjustl(fn_1))//'-np'//trim(adjustl(c_ts))//'.dat' |
---|
| 2893 | |
---|
| 2894 | call get_lun(dat_lun) |
---|
| 2895 | open(UNIT=dat_lun, FILE=fn_2, IOSTAT=lios, ACTION='WRITE', & |
---|
| 2896 | FORM='UNFORMATTED', STATUS='REPLACE') |
---|
| 2897 | write(UNIT=dat_lun, IOSTAT=lios) array_in |
---|
| 2898 | close(UNIT=dat_lun) |
---|
| 2899 | |
---|
| 2900 | end subroutine write_data_dbg |
---|
| 2901 | |
---|
| 2902 | |
---|
| 2903 | subroutine set_fields_synthetic() |
---|
| 2904 | !******************************************************************************* |
---|
| 2905 | ! DESCRIPTION |
---|
| 2906 | ! Set all meteorological fields to synthetic (usually constant/homogeneous) |
---|
| 2907 | ! values. |
---|
| 2908 | ! Used for validation and error-checking |
---|
| 2909 | ! |
---|
| 2910 | ! NOTE |
---|
| 2911 | ! This version uses asynchronious communications. |
---|
| 2912 | ! |
---|
| 2913 | ! VARIABLES |
---|
| 2914 | ! |
---|
| 2915 | ! |
---|
| 2916 | ! |
---|
| 2917 | !******************************************************************************* |
---|
| 2918 | use com_mod |
---|
| 2919 | |
---|
| 2920 | implicit none |
---|
[8a65cb0] | 2921 | |
---|
[861805a] | 2922 | integer :: li=1, ui=2 ! wfmem indices (i.e, operate on entire field) |
---|
| 2923 | |
---|
| 2924 | if (.not.lmp_sync) ui=3 |
---|
| 2925 | |
---|
| 2926 | |
---|
| 2927 | ! Variables transferred at initialization only |
---|
| 2928 | !********************************************* |
---|
| 2929 | ! readclouds=readclouds_ |
---|
| 2930 | oro(:,:)=0.0 |
---|
| 2931 | excessoro(:,:)=0.0 |
---|
| 2932 | lsm(:,:)=0.0 |
---|
| 2933 | xlanduse(:,:,:)=0.0 |
---|
| 2934 | ! wftime |
---|
| 2935 | ! numbwf |
---|
| 2936 | ! nmixz |
---|
| 2937 | ! height |
---|
| 2938 | |
---|
| 2939 | ! Time-varying fields: |
---|
| 2940 | uu(:,:,:,li:ui) = 10.0 |
---|
| 2941 | vv(:,:,:,li:ui) = 0.0 |
---|
| 2942 | uupol(:,:,:,li:ui) = 10.0 |
---|
| 2943 | vvpol(:,:,:,li:ui)=0.0 |
---|
| 2944 | ww(:,:,:,li:ui)=0. |
---|
| 2945 | tt(:,:,:,li:ui)=300. |
---|
| 2946 | rho(:,:,:,li:ui)=1.3 |
---|
| 2947 | drhodz(:,:,:,li:ui)=.0 |
---|
| 2948 | tth(:,:,:,li:ui)=0.0 |
---|
| 2949 | qvh(:,:,:,li:ui)=1.0 |
---|
| 2950 | qv(:,:,:,li:ui)=1.0 |
---|
| 2951 | |
---|
| 2952 | pv(:,:,:,li:ui)=1.0 |
---|
| 2953 | clouds(:,:,:,li:ui)=0 |
---|
| 2954 | |
---|
| 2955 | clwc(:,:,:,li:ui)=0.0 |
---|
| 2956 | ciwc(:,:,:,li:ui)=0.0 |
---|
| 2957 | |
---|
| 2958 | ! 2D fields |
---|
[8a65cb0] | 2959 | |
---|
[861805a] | 2960 | cloudsh(:,:,li:ui)=0 |
---|
| 2961 | vdep(:,:,:,li:ui)=0.0 |
---|
| 2962 | ps(:,:,:,li:ui)=1.0e5 |
---|
| 2963 | sd(:,:,:,li:ui)=0.0 |
---|
| 2964 | tcc(:,:,:,li:ui)=0.0 |
---|
| 2965 | tt2(:,:,:,li:ui)=300. |
---|
| 2966 | td2(:,:,:,li:ui)=250. |
---|
| 2967 | lsprec(:,:,:,li:ui)=0.0 |
---|
| 2968 | convprec(:,:,:,li:ui)=0.0 |
---|
| 2969 | ustar(:,:,:,li:ui)=1.0 |
---|
| 2970 | wstar(:,:,:,li:ui)=1.0 |
---|
| 2971 | hmix(:,:,:,li:ui)=10000. |
---|
| 2972 | tropopause(:,:,:,li:ui)=10000. |
---|
| 2973 | oli(:,:,:,li:ui)=0.01 |
---|
| 2974 | |
---|
| 2975 | end subroutine set_fields_synthetic |
---|
[8a65cb0] | 2976 | |
---|
| 2977 | end module mpi_mod |
---|