[1216] | 1 | SUBROUTINE resort_for_xy( f_in, f_inv ) |
---|
[1] | 2 | |
---|
[1036] | 3 | !--------------------------------------------------------------------------------! |
---|
| 4 | ! This file is part of PALM. |
---|
| 5 | ! |
---|
| 6 | ! PALM is free software: you can redistribute it and/or modify it under the terms |
---|
| 7 | ! of the GNU General Public License as published by the Free Software Foundation, |
---|
| 8 | ! either version 3 of the License, or (at your option) any later version. |
---|
| 9 | ! |
---|
| 10 | ! PALM is distributed in the hope that it will be useful, but WITHOUT ANY |
---|
| 11 | ! WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR |
---|
| 12 | ! A PARTICULAR PURPOSE. See the GNU General Public License for more details. |
---|
| 13 | ! |
---|
| 14 | ! You should have received a copy of the GNU General Public License along with |
---|
| 15 | ! PALM. If not, see <http://www.gnu.org/licenses/>. |
---|
| 16 | ! |
---|
| 17 | ! Copyright 1997-2012 Leibniz University Hannover |
---|
| 18 | !--------------------------------------------------------------------------------! |
---|
| 19 | ! |
---|
[484] | 20 | ! Current revisions: |
---|
[1] | 21 | ! ----------------- |
---|
[198] | 22 | ! |
---|
[1258] | 23 | ! |
---|
[198] | 24 | ! Former revisions: |
---|
| 25 | ! ----------------- |
---|
| 26 | ! $Id: transpose.f90 1258 2013-11-08 16:09:09Z heinze $ |
---|
| 27 | ! |
---|
[1258] | 28 | ! 1257 2013-11-08 15:18:40Z raasch |
---|
| 29 | ! openacc loop and loop vector clauses removed |
---|
| 30 | ! |
---|
[1217] | 31 | ! 1216 2013-08-26 09:31:42Z raasch |
---|
| 32 | ! re-sorting of the transposed / to be transposed arrays moved to separate |
---|
| 33 | ! routines resort_for_... |
---|
| 34 | ! |
---|
[1112] | 35 | ! 1111 2013-03-08 23:54:10Z raasch |
---|
| 36 | ! openACC directives added, |
---|
| 37 | ! resorting data from/to work changed, work got 4 dimensions instead of 1 |
---|
| 38 | ! |
---|
[1107] | 39 | ! 1106 2013-03-04 05:31:38Z raasch |
---|
| 40 | ! preprocessor lines rearranged so that routines can also be used in serial |
---|
| 41 | ! (non-parallel) mode |
---|
| 42 | ! |
---|
[1093] | 43 | ! 1092 2013-02-02 11:24:22Z raasch |
---|
| 44 | ! unused variables removed |
---|
| 45 | ! |
---|
[1037] | 46 | ! 1036 2012-10-22 13:43:42Z raasch |
---|
| 47 | ! code put under GPL (PALM 3.9) |
---|
| 48 | ! |
---|
[1004] | 49 | ! 1003 2012-09-14 14:35:53Z raasch |
---|
| 50 | ! indices nxa, nya, etc. replaced by nx, ny, etc. |
---|
| 51 | ! |
---|
[684] | 52 | ! 683 2011-02-09 14:25:15Z raasch |
---|
| 53 | ! openMP parallelization of transpositions for 2d-domain-decomposition |
---|
| 54 | ! |
---|
[623] | 55 | ! 622 2010-12-10 08:08:13Z raasch |
---|
| 56 | ! optional barriers included in order to speed up collective operations |
---|
| 57 | ! |
---|
[198] | 58 | ! 164 2008-05-15 08:46:15Z raasch |
---|
[164] | 59 | ! f_inv changed from subroutine argument to automatic array in order to do |
---|
| 60 | ! re-ordering from f_in to f_inv in one step, one array work is needed instead |
---|
| 61 | ! of work1 and work2 |
---|
[1] | 62 | ! |
---|
[198] | 63 | ! February 2007 |
---|
[3] | 64 | ! RCS Log replace by Id keyword, revision history cleaned up |
---|
| 65 | ! |
---|
[1] | 66 | ! Revision 1.2 2004/04/30 13:12:17 raasch |
---|
| 67 | ! Switched from mpi_alltoallv to the simpler mpi_alltoall, |
---|
| 68 | ! all former transpose-routine files collected in this file, enlarged |
---|
| 69 | ! transposition arrays introduced |
---|
| 70 | ! |
---|
| 71 | ! Revision 1.1 2004/04/30 13:08:16 raasch |
---|
| 72 | ! Initial revision (collection of former routines transpose_xy, transpose_xz, |
---|
| 73 | ! transpose_yx, transpose_yz, transpose_zx, transpose_zy) |
---|
| 74 | ! |
---|
| 75 | ! Revision 1.1 1997/07/24 11:25:18 raasch |
---|
| 76 | ! Initial revision |
---|
| 77 | ! |
---|
[1216] | 78 | !------------------------------------------------------------------------------! |
---|
| 79 | ! Description: |
---|
| 80 | ! ------------ |
---|
| 81 | ! Resorting data for the transposition from x to y. The transposition itself |
---|
| 82 | ! is carried out in transpose_xy |
---|
| 83 | !------------------------------------------------------------------------------! |
---|
| 84 | |
---|
| 85 | USE indices |
---|
| 86 | USE transpose_indices |
---|
| 87 | |
---|
| 88 | IMPLICIT NONE |
---|
| 89 | |
---|
| 90 | REAL :: f_in(0:nx,nys_x:nyn_x,nzb_x:nzt_x) |
---|
| 91 | REAL :: f_inv(nys_x:nyn_x,nzb_x:nzt_x,0:nx) |
---|
| 92 | |
---|
| 93 | |
---|
| 94 | INTEGER :: i, j, k |
---|
| 95 | |
---|
[1] | 96 | ! |
---|
[1216] | 97 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 98 | !-- by MPI contiguous |
---|
| 99 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 100 | !$OMP DO |
---|
| 101 | !$acc kernels present( f_in, f_inv ) |
---|
| 102 | DO i = 0, nx |
---|
| 103 | DO k = nzb_x, nzt_x |
---|
| 104 | DO j = nys_x, nyn_x |
---|
| 105 | f_inv(j,k,i) = f_in(i,j,k) |
---|
| 106 | ENDDO |
---|
| 107 | ENDDO |
---|
| 108 | ENDDO |
---|
| 109 | !$acc end kernels |
---|
| 110 | !$OMP END PARALLEL |
---|
| 111 | |
---|
| 112 | END SUBROUTINE resort_for_xy |
---|
| 113 | |
---|
| 114 | |
---|
| 115 | SUBROUTINE transpose_xy( f_inv, f_out ) |
---|
| 116 | |
---|
| 117 | !------------------------------------------------------------------------------! |
---|
[1] | 118 | ! Description: |
---|
| 119 | ! ------------ |
---|
| 120 | ! Transposition of input array (f_in) from x to y. For the input array, all |
---|
| 121 | ! elements along x reside on the same PE, while after transposition, all |
---|
| 122 | ! elements along y reside on the same PE. |
---|
| 123 | !------------------------------------------------------------------------------! |
---|
| 124 | |
---|
| 125 | USE cpulog |
---|
| 126 | USE indices |
---|
| 127 | USE interfaces |
---|
| 128 | USE pegrid |
---|
| 129 | USE transpose_indices |
---|
| 130 | |
---|
| 131 | IMPLICIT NONE |
---|
| 132 | |
---|
[1111] | 133 | INTEGER :: i, j, k, l, ys |
---|
[1] | 134 | |
---|
[1216] | 135 | REAL :: f_inv(nys_x:nyn_x,nzb_x:nzt_x,0:nx), f_out(0:ny,nxl_y:nxr_y,nzb_y:nzt_y) |
---|
[1] | 136 | |
---|
[1111] | 137 | REAL, DIMENSION(nyn_x-nys_x+1,nzb_y:nzt_y,nxl_y:nxr_y,0:pdims(2)-1) :: work |
---|
| 138 | |
---|
| 139 | |
---|
[1106] | 140 | IF ( numprocs /= 1 ) THEN |
---|
| 141 | |
---|
| 142 | #if defined( __parallel ) |
---|
[1] | 143 | ! |
---|
[1106] | 144 | !-- Transpose array |
---|
| 145 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
| 146 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1111] | 147 | !$acc update host( f_inv ) |
---|
| 148 | CALL MPI_ALLTOALL( f_inv(nys_x,nzb_x,0), sendrecvcount_xy, MPI_REAL, & |
---|
| 149 | work(1,nzb_y,nxl_y,0), sendrecvcount_xy, MPI_REAL, & |
---|
[1106] | 150 | comm1dy, ierr ) |
---|
| 151 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
[1] | 152 | |
---|
| 153 | ! |
---|
[1106] | 154 | !-- Reorder transposed array |
---|
[1111] | 155 | !$OMP PARALLEL PRIVATE ( i, j, k, l, ys ) |
---|
[683] | 156 | !$OMP DO |
---|
[1216] | 157 | !$acc data copyin( work ) |
---|
[1106] | 158 | DO l = 0, pdims(2) - 1 |
---|
| 159 | ys = 0 + l * ( nyn_x - nys_x + 1 ) |
---|
[1111] | 160 | !$acc kernels present( f_out, work ) |
---|
[1106] | 161 | DO i = nxl_y, nxr_y |
---|
| 162 | DO k = nzb_y, nzt_y |
---|
| 163 | DO j = ys, ys + nyn_x - nys_x |
---|
[1111] | 164 | f_out(j,i,k) = work(j-ys+1,k,i,l) |
---|
[1106] | 165 | ENDDO |
---|
[1] | 166 | ENDDO |
---|
| 167 | ENDDO |
---|
[1111] | 168 | !$acc end kernels |
---|
[1] | 169 | ENDDO |
---|
[1216] | 170 | !$acc end data |
---|
[683] | 171 | !$OMP END PARALLEL |
---|
[1] | 172 | #endif |
---|
| 173 | |
---|
[1106] | 174 | ELSE |
---|
| 175 | |
---|
| 176 | ! |
---|
| 177 | !-- Reorder transposed array |
---|
| 178 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 179 | !$OMP DO |
---|
[1216] | 180 | !$acc kernels present( f_inv, f_out ) |
---|
[1106] | 181 | DO k = nzb_y, nzt_y |
---|
| 182 | DO i = nxl_y, nxr_y |
---|
| 183 | DO j = 0, ny |
---|
| 184 | f_out(j,i,k) = f_inv(j,k,i) |
---|
| 185 | ENDDO |
---|
| 186 | ENDDO |
---|
| 187 | ENDDO |
---|
[1111] | 188 | !$acc end kernels |
---|
[1106] | 189 | !$OMP END PARALLEL |
---|
| 190 | |
---|
| 191 | ENDIF |
---|
| 192 | |
---|
[1] | 193 | END SUBROUTINE transpose_xy |
---|
| 194 | |
---|
| 195 | |
---|
[1216] | 196 | SUBROUTINE resort_for_xz( f_inv, f_out ) |
---|
[1] | 197 | |
---|
| 198 | !------------------------------------------------------------------------------! |
---|
| 199 | ! Description: |
---|
| 200 | ! ------------ |
---|
[1216] | 201 | ! Resorting data after the transposition from x to z. The transposition itself |
---|
| 202 | ! is carried out in transpose_xz |
---|
| 203 | !------------------------------------------------------------------------------! |
---|
| 204 | |
---|
| 205 | USE indices |
---|
| 206 | USE transpose_indices |
---|
| 207 | |
---|
| 208 | IMPLICIT NONE |
---|
| 209 | |
---|
| 210 | REAL :: f_inv(nys:nyn,nxl:nxr,1:nz) |
---|
| 211 | REAL :: f_out(1:nz,nys:nyn,nxl:nxr) |
---|
| 212 | |
---|
| 213 | |
---|
| 214 | INTEGER :: i, j, k |
---|
| 215 | |
---|
| 216 | ! |
---|
| 217 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 218 | !-- by MPI contiguous. |
---|
| 219 | !-- In case of parallel fft/transposition, scattered store is faster in |
---|
| 220 | !-- backward direction!!! |
---|
| 221 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 222 | !$OMP DO |
---|
| 223 | !$acc kernels present( f_inv, f_out ) |
---|
| 224 | DO k = 1, nz |
---|
| 225 | DO i = nxl, nxr |
---|
| 226 | DO j = nys, nyn |
---|
| 227 | f_out(k,j,i) = f_inv(j,i,k) |
---|
| 228 | ENDDO |
---|
| 229 | ENDDO |
---|
| 230 | ENDDO |
---|
| 231 | !$acc end kernels |
---|
| 232 | !$OMP END PARALLEL |
---|
| 233 | |
---|
| 234 | END SUBROUTINE resort_for_xz |
---|
| 235 | |
---|
| 236 | |
---|
| 237 | SUBROUTINE transpose_xz( f_in, f_inv ) |
---|
| 238 | |
---|
| 239 | !------------------------------------------------------------------------------! |
---|
| 240 | ! Description: |
---|
| 241 | ! ------------ |
---|
[1] | 242 | ! Transposition of input array (f_in) from x to z. For the input array, all |
---|
| 243 | ! elements along x reside on the same PE, while after transposition, all |
---|
| 244 | ! elements along z reside on the same PE. |
---|
| 245 | !------------------------------------------------------------------------------! |
---|
| 246 | |
---|
| 247 | USE cpulog |
---|
| 248 | USE indices |
---|
| 249 | USE interfaces |
---|
| 250 | USE pegrid |
---|
| 251 | USE transpose_indices |
---|
| 252 | |
---|
| 253 | IMPLICIT NONE |
---|
| 254 | |
---|
[1111] | 255 | INTEGER :: i, j, k, l, xs |
---|
[1] | 256 | |
---|
[1216] | 257 | REAL :: f_in(0:nx,nys_x:nyn_x,nzb_x:nzt_x), f_inv(nys:nyn,nxl:nxr,1:nz) |
---|
[1] | 258 | |
---|
[1111] | 259 | REAL, DIMENSION(nys_x:nyn_x,nnx,nzb_x:nzt_x,0:pdims(1)-1) :: work |
---|
[1] | 260 | |
---|
[1111] | 261 | |
---|
[1] | 262 | ! |
---|
| 263 | !-- If the PE grid is one-dimensional along y, the array has only to be |
---|
| 264 | !-- reordered locally and therefore no transposition has to be done. |
---|
| 265 | IF ( pdims(1) /= 1 ) THEN |
---|
[1106] | 266 | |
---|
| 267 | #if defined( __parallel ) |
---|
[1] | 268 | ! |
---|
| 269 | !-- Reorder input array for transposition |
---|
[1111] | 270 | !$OMP PARALLEL PRIVATE ( i, j, k, l, xs ) |
---|
[683] | 271 | !$OMP DO |
---|
[1216] | 272 | !$acc data copyout( work ) |
---|
[1] | 273 | DO l = 0, pdims(1) - 1 |
---|
| 274 | xs = 0 + l * nnx |
---|
[1111] | 275 | !$acc kernels present( f_in, work ) |
---|
[1003] | 276 | DO k = nzb_x, nzt_x |
---|
[164] | 277 | DO i = xs, xs + nnx - 1 |
---|
[1003] | 278 | DO j = nys_x, nyn_x |
---|
[1111] | 279 | work(j,i-xs+1,k,l) = f_in(i,j,k) |
---|
[1] | 280 | ENDDO |
---|
| 281 | ENDDO |
---|
| 282 | ENDDO |
---|
[1111] | 283 | !$acc end kernels |
---|
[1] | 284 | ENDDO |
---|
[1216] | 285 | !$acc end data |
---|
[683] | 286 | !$OMP END PARALLEL |
---|
[1] | 287 | |
---|
| 288 | ! |
---|
| 289 | !-- Transpose array |
---|
| 290 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
[622] | 291 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1111] | 292 | CALL MPI_ALLTOALL( work(nys_x,1,nzb_x,0), sendrecvcount_zx, MPI_REAL, & |
---|
| 293 | f_inv(nys,nxl,1), sendrecvcount_zx, MPI_REAL, & |
---|
[1] | 294 | comm1dx, ierr ) |
---|
[1111] | 295 | !$acc update device( f_inv ) |
---|
[1] | 296 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
[1106] | 297 | #endif |
---|
| 298 | |
---|
[1] | 299 | ELSE |
---|
[1106] | 300 | |
---|
[1] | 301 | ! |
---|
| 302 | !-- Reorder the array in a way that the z index is in first position |
---|
[683] | 303 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 304 | !$OMP DO |
---|
[1216] | 305 | !$acc kernels present( f_in, f_inv ) |
---|
[1003] | 306 | DO i = nxl, nxr |
---|
| 307 | DO j = nys, nyn |
---|
| 308 | DO k = 1, nz |
---|
[164] | 309 | f_inv(j,i,k) = f_in(i,j,k) |
---|
[1] | 310 | ENDDO |
---|
| 311 | ENDDO |
---|
| 312 | ENDDO |
---|
[1111] | 313 | !$acc end kernels |
---|
[683] | 314 | !$OMP END PARALLEL |
---|
[1] | 315 | |
---|
[164] | 316 | ENDIF |
---|
| 317 | |
---|
[1] | 318 | END SUBROUTINE transpose_xz |
---|
| 319 | |
---|
| 320 | |
---|
[1216] | 321 | SUBROUTINE resort_for_yx( f_inv, f_out ) |
---|
[1] | 322 | |
---|
| 323 | !------------------------------------------------------------------------------! |
---|
| 324 | ! Description: |
---|
| 325 | ! ------------ |
---|
[1216] | 326 | ! Resorting data after the transposition from y to x. The transposition itself |
---|
| 327 | ! is carried out in transpose_yx |
---|
| 328 | !------------------------------------------------------------------------------! |
---|
| 329 | |
---|
| 330 | USE indices |
---|
| 331 | USE transpose_indices |
---|
| 332 | |
---|
| 333 | IMPLICIT NONE |
---|
| 334 | |
---|
| 335 | REAL :: f_inv(nys_x:nyn_x,nzb_x:nzt_x,0:nx) |
---|
| 336 | REAL :: f_out(0:nx,nys_x:nyn_x,nzb_x:nzt_x) |
---|
| 337 | |
---|
| 338 | |
---|
| 339 | INTEGER :: i, j, k |
---|
| 340 | |
---|
| 341 | ! |
---|
| 342 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 343 | !-- by MPI contiguous |
---|
| 344 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 345 | !$OMP DO |
---|
| 346 | !$acc kernels present( f_inv, f_out ) |
---|
| 347 | DO i = 0, nx |
---|
| 348 | DO k = nzb_x, nzt_x |
---|
| 349 | DO j = nys_x, nyn_x |
---|
| 350 | f_out(i,j,k) = f_inv(j,k,i) |
---|
| 351 | ENDDO |
---|
| 352 | ENDDO |
---|
| 353 | ENDDO |
---|
| 354 | !$acc end kernels |
---|
| 355 | !$OMP END PARALLEL |
---|
| 356 | |
---|
| 357 | END SUBROUTINE resort_for_yx |
---|
| 358 | |
---|
| 359 | |
---|
| 360 | SUBROUTINE transpose_yx( f_in, f_inv ) |
---|
| 361 | |
---|
| 362 | !------------------------------------------------------------------------------! |
---|
| 363 | ! Description: |
---|
| 364 | ! ------------ |
---|
[1] | 365 | ! Transposition of input array (f_in) from y to x. For the input array, all |
---|
| 366 | ! elements along y reside on the same PE, while after transposition, all |
---|
| 367 | ! elements along x reside on the same PE. |
---|
| 368 | !------------------------------------------------------------------------------! |
---|
| 369 | |
---|
| 370 | USE cpulog |
---|
| 371 | USE indices |
---|
| 372 | USE interfaces |
---|
| 373 | USE pegrid |
---|
| 374 | USE transpose_indices |
---|
| 375 | |
---|
| 376 | IMPLICIT NONE |
---|
| 377 | |
---|
[1111] | 378 | INTEGER :: i, j, k, l, ys |
---|
[1] | 379 | |
---|
[1216] | 380 | REAL :: f_in(0:ny,nxl_y:nxr_y,nzb_y:nzt_y), f_inv(nys_x:nyn_x,nzb_x:nzt_x,0:nx) |
---|
[1] | 381 | |
---|
[1111] | 382 | REAL, DIMENSION(nyn_x-nys_x+1,nzb_y:nzt_y,nxl_y:nxr_y,0:pdims(2)-1) :: work |
---|
| 383 | |
---|
| 384 | |
---|
[1106] | 385 | IF ( numprocs /= 1 ) THEN |
---|
| 386 | |
---|
[1] | 387 | #if defined( __parallel ) |
---|
| 388 | ! |
---|
[1106] | 389 | !-- Reorder input array for transposition |
---|
[1111] | 390 | !$OMP PARALLEL PRIVATE ( i, j, k, l, ys ) |
---|
[683] | 391 | !$OMP DO |
---|
[1216] | 392 | !$acc data copyout( work ) |
---|
[1106] | 393 | DO l = 0, pdims(2) - 1 |
---|
| 394 | ys = 0 + l * ( nyn_x - nys_x + 1 ) |
---|
[1111] | 395 | !$acc kernels present( f_in, work ) |
---|
[1106] | 396 | DO i = nxl_y, nxr_y |
---|
| 397 | DO k = nzb_y, nzt_y |
---|
| 398 | DO j = ys, ys + nyn_x - nys_x |
---|
[1111] | 399 | work(j-ys+1,k,i,l) = f_in(j,i,k) |
---|
[1106] | 400 | ENDDO |
---|
| 401 | ENDDO |
---|
| 402 | ENDDO |
---|
[1111] | 403 | !$acc end kernels |
---|
[1106] | 404 | ENDDO |
---|
[1216] | 405 | !$acc end data |
---|
[1106] | 406 | !$OMP END PARALLEL |
---|
| 407 | |
---|
| 408 | ! |
---|
| 409 | !-- Transpose array |
---|
| 410 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
| 411 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1111] | 412 | CALL MPI_ALLTOALL( work(1,nzb_y,nxl_y,0), sendrecvcount_xy, MPI_REAL, & |
---|
| 413 | f_inv(nys_x,nzb_x,0), sendrecvcount_xy, MPI_REAL, & |
---|
[1106] | 414 | comm1dy, ierr ) |
---|
[1111] | 415 | !$acc update device( f_inv ) |
---|
[1106] | 416 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
| 417 | #endif |
---|
| 418 | |
---|
| 419 | ELSE |
---|
| 420 | |
---|
| 421 | ! |
---|
| 422 | !-- Reorder array f_in the same way as ALLTOALL did it |
---|
| 423 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 424 | !$OMP DO |
---|
[1216] | 425 | !$acc kernels present( f_in, f_inv ) |
---|
[1003] | 426 | DO i = nxl_y, nxr_y |
---|
| 427 | DO k = nzb_y, nzt_y |
---|
[1106] | 428 | DO j = 0, ny |
---|
| 429 | f_inv(j,k,i) = f_in(j,i,k) |
---|
[1] | 430 | ENDDO |
---|
| 431 | ENDDO |
---|
| 432 | ENDDO |
---|
[1111] | 433 | !$acc end kernels |
---|
[683] | 434 | !$OMP END PARALLEL |
---|
[1] | 435 | |
---|
[1106] | 436 | ENDIF |
---|
[1] | 437 | |
---|
| 438 | END SUBROUTINE transpose_yx |
---|
| 439 | |
---|
| 440 | |
---|
[1216] | 441 | SUBROUTINE transpose_yxd( f_in, f_out ) |
---|
[1] | 442 | |
---|
| 443 | !------------------------------------------------------------------------------! |
---|
| 444 | ! Description: |
---|
| 445 | ! ------------ |
---|
| 446 | ! Transposition of input array (f_in) from y to x. For the input array, all |
---|
| 447 | ! elements along y reside on the same PE, while after transposition, all |
---|
| 448 | ! elements along x reside on the same PE. |
---|
| 449 | ! This is a direct transposition for arrays with indices in regular order |
---|
| 450 | ! (k,j,i) (cf. transpose_yx). |
---|
| 451 | !------------------------------------------------------------------------------! |
---|
| 452 | |
---|
| 453 | USE cpulog |
---|
| 454 | USE indices |
---|
| 455 | USE interfaces |
---|
| 456 | USE pegrid |
---|
| 457 | USE transpose_indices |
---|
| 458 | |
---|
| 459 | IMPLICIT NONE |
---|
| 460 | |
---|
[1092] | 461 | INTEGER :: i, j, k, l, m, xs |
---|
[1] | 462 | |
---|
[1003] | 463 | REAL :: f_in(1:nz,nys:nyn,nxl:nxr), f_inv(nxl:nxr,1:nz,nys:nyn), & |
---|
| 464 | f_out(0:nx,nys_x:nyn_x,nzb_x:nzt_x), & |
---|
[164] | 465 | work(nnx*nny*nnz) |
---|
[1] | 466 | |
---|
| 467 | #if defined( __parallel ) |
---|
| 468 | |
---|
| 469 | ! |
---|
| 470 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 471 | !-- by MPI contiguous |
---|
[1003] | 472 | DO k = 1, nz |
---|
| 473 | DO j = nys, nyn |
---|
| 474 | DO i = nxl, nxr |
---|
[164] | 475 | f_inv(i,k,j) = f_in(k,j,i) |
---|
[1] | 476 | ENDDO |
---|
| 477 | ENDDO |
---|
| 478 | ENDDO |
---|
| 479 | |
---|
| 480 | ! |
---|
| 481 | !-- Transpose array |
---|
| 482 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
[622] | 483 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1] | 484 | CALL MPI_ALLTOALL( f_inv(nxl,1,nys), sendrecvcount_xy, MPI_REAL, & |
---|
[164] | 485 | work(1), sendrecvcount_xy, MPI_REAL, & |
---|
[1] | 486 | comm1dx, ierr ) |
---|
| 487 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
| 488 | |
---|
| 489 | ! |
---|
| 490 | !-- Reorder transposed array |
---|
| 491 | m = 0 |
---|
| 492 | DO l = 0, pdims(1) - 1 |
---|
| 493 | xs = 0 + l * nnx |
---|
[1003] | 494 | DO j = nys_x, nyn_x |
---|
| 495 | DO k = 1, nz |
---|
[1] | 496 | DO i = xs, xs + nnx - 1 |
---|
| 497 | m = m + 1 |
---|
[164] | 498 | f_out(i,j,k) = work(m) |
---|
[1] | 499 | ENDDO |
---|
| 500 | ENDDO |
---|
| 501 | ENDDO |
---|
| 502 | ENDDO |
---|
| 503 | |
---|
| 504 | #endif |
---|
| 505 | |
---|
| 506 | END SUBROUTINE transpose_yxd |
---|
| 507 | |
---|
| 508 | |
---|
[1216] | 509 | SUBROUTINE resort_for_yz( f_in, f_inv ) |
---|
[1] | 510 | |
---|
| 511 | !------------------------------------------------------------------------------! |
---|
| 512 | ! Description: |
---|
| 513 | ! ------------ |
---|
[1216] | 514 | ! Resorting data for the transposition from y to z. The transposition itself |
---|
| 515 | ! is carried out in transpose_yz |
---|
| 516 | !------------------------------------------------------------------------------! |
---|
| 517 | |
---|
| 518 | USE indices |
---|
| 519 | USE transpose_indices |
---|
| 520 | |
---|
| 521 | IMPLICIT NONE |
---|
| 522 | |
---|
| 523 | REAL :: f_in(0:ny,nxl_y:nxr_y,nzb_y:nzt_y) |
---|
| 524 | REAL :: f_inv(nxl_y:nxr_y,nzb_y:nzt_y,0:ny) |
---|
| 525 | |
---|
| 526 | |
---|
| 527 | INTEGER :: i, j, k |
---|
| 528 | |
---|
| 529 | ! |
---|
| 530 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 531 | !-- by MPI contiguous |
---|
| 532 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 533 | !$OMP DO |
---|
| 534 | !$acc kernels present( f_in, f_inv ) |
---|
| 535 | DO j = 0, ny |
---|
| 536 | DO k = nzb_y, nzt_y |
---|
| 537 | DO i = nxl_y, nxr_y |
---|
| 538 | f_inv(i,k,j) = f_in(j,i,k) |
---|
| 539 | ENDDO |
---|
| 540 | ENDDO |
---|
| 541 | ENDDO |
---|
| 542 | !$acc end kernels |
---|
| 543 | !$OMP END PARALLEL |
---|
| 544 | |
---|
| 545 | END SUBROUTINE resort_for_yz |
---|
| 546 | |
---|
| 547 | |
---|
| 548 | SUBROUTINE transpose_yz( f_inv, f_out ) |
---|
| 549 | |
---|
| 550 | !------------------------------------------------------------------------------! |
---|
| 551 | ! Description: |
---|
| 552 | ! ------------ |
---|
[1] | 553 | ! Transposition of input array (f_in) from y to z. For the input array, all |
---|
| 554 | ! elements along y reside on the same PE, while after transposition, all |
---|
| 555 | ! elements along z reside on the same PE. |
---|
| 556 | !------------------------------------------------------------------------------! |
---|
| 557 | |
---|
| 558 | USE cpulog |
---|
| 559 | USE indices |
---|
| 560 | USE interfaces |
---|
| 561 | USE pegrid |
---|
| 562 | USE transpose_indices |
---|
| 563 | |
---|
| 564 | IMPLICIT NONE |
---|
| 565 | |
---|
[1111] | 566 | INTEGER :: i, j, k, l, zs |
---|
[1] | 567 | |
---|
[1216] | 568 | REAL :: f_inv(nxl_y:nxr_y,nzb_y:nzt_y,0:ny), f_out(nxl_z:nxr_z,nys_z:nyn_z,1:nz) |
---|
[1] | 569 | |
---|
[1111] | 570 | REAL, DIMENSION(nxl_z:nxr_z,nzt_y-nzb_y+1,nys_z:nyn_z,0:pdims(1)-1) :: work |
---|
| 571 | |
---|
| 572 | |
---|
[1] | 573 | ! |
---|
| 574 | !-- If the PE grid is one-dimensional along y, only local reordering |
---|
| 575 | !-- of the data is necessary and no transposition has to be done. |
---|
| 576 | IF ( pdims(1) == 1 ) THEN |
---|
[1106] | 577 | |
---|
[683] | 578 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 579 | !$OMP DO |
---|
[1216] | 580 | !$acc kernels present( f_inv, f_out ) |
---|
[1003] | 581 | DO j = 0, ny |
---|
| 582 | DO k = nzb_y, nzt_y |
---|
| 583 | DO i = nxl_y, nxr_y |
---|
[164] | 584 | f_out(i,j,k) = f_inv(i,k,j) |
---|
[1] | 585 | ENDDO |
---|
| 586 | ENDDO |
---|
| 587 | ENDDO |
---|
[1111] | 588 | !$acc end kernels |
---|
[683] | 589 | !$OMP END PARALLEL |
---|
[1] | 590 | |
---|
[1106] | 591 | ELSE |
---|
| 592 | |
---|
| 593 | #if defined( __parallel ) |
---|
[1] | 594 | ! |
---|
[1106] | 595 | !-- Transpose array |
---|
| 596 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
| 597 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1111] | 598 | !$acc update host( f_inv ) |
---|
| 599 | CALL MPI_ALLTOALL( f_inv(nxl_y,nzb_y,0), sendrecvcount_yz, MPI_REAL, & |
---|
| 600 | work(nxl_z,1,nys_z,0), sendrecvcount_yz, MPI_REAL, & |
---|
[1106] | 601 | comm1dx, ierr ) |
---|
| 602 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
[1] | 603 | |
---|
| 604 | ! |
---|
[1106] | 605 | !-- Reorder transposed array |
---|
[1111] | 606 | !$OMP PARALLEL PRIVATE ( i, j, k, l, zs ) |
---|
[683] | 607 | !$OMP DO |
---|
[1216] | 608 | !$acc data copyin( work ) |
---|
[1106] | 609 | DO l = 0, pdims(1) - 1 |
---|
| 610 | zs = 1 + l * ( nzt_y - nzb_y + 1 ) |
---|
[1216] | 611 | !$acc kernels present( f_out ) |
---|
[1106] | 612 | DO j = nys_z, nyn_z |
---|
| 613 | DO k = zs, zs + nzt_y - nzb_y |
---|
| 614 | DO i = nxl_z, nxr_z |
---|
[1111] | 615 | f_out(i,j,k) = work(i,k-zs+1,j,l) |
---|
[1106] | 616 | ENDDO |
---|
[1] | 617 | ENDDO |
---|
| 618 | ENDDO |
---|
[1111] | 619 | !$acc end kernels |
---|
[1] | 620 | ENDDO |
---|
[1216] | 621 | !$acc end data |
---|
[683] | 622 | !$OMP END PARALLEL |
---|
[1] | 623 | #endif |
---|
| 624 | |
---|
[1106] | 625 | ENDIF |
---|
| 626 | |
---|
[1] | 627 | END SUBROUTINE transpose_yz |
---|
| 628 | |
---|
| 629 | |
---|
[1216] | 630 | SUBROUTINE resort_for_zx( f_in, f_inv ) |
---|
[1] | 631 | |
---|
| 632 | !------------------------------------------------------------------------------! |
---|
| 633 | ! Description: |
---|
| 634 | ! ------------ |
---|
[1216] | 635 | ! Resorting data for the transposition from z to x. The transposition itself |
---|
| 636 | ! is carried out in transpose_zx |
---|
| 637 | !------------------------------------------------------------------------------! |
---|
| 638 | |
---|
| 639 | USE indices |
---|
| 640 | USE transpose_indices |
---|
| 641 | |
---|
| 642 | IMPLICIT NONE |
---|
| 643 | |
---|
| 644 | REAL :: f_in(1:nz,nys:nyn,nxl:nxr) |
---|
| 645 | REAL :: f_inv(nys:nyn,nxl:nxr,1:nz) |
---|
| 646 | |
---|
| 647 | |
---|
| 648 | INTEGER :: i, j, k |
---|
| 649 | |
---|
| 650 | ! |
---|
| 651 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 652 | !-- by MPI contiguous |
---|
| 653 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 654 | !$OMP DO |
---|
| 655 | !$acc kernels present( f_in, f_inv ) |
---|
| 656 | DO k = 1,nz |
---|
| 657 | DO i = nxl, nxr |
---|
| 658 | DO j = nys, nyn |
---|
| 659 | f_inv(j,i,k) = f_in(k,j,i) |
---|
| 660 | ENDDO |
---|
| 661 | ENDDO |
---|
| 662 | ENDDO |
---|
| 663 | !$acc end kernels |
---|
| 664 | !$OMP END PARALLEL |
---|
| 665 | |
---|
| 666 | END SUBROUTINE resort_for_zx |
---|
| 667 | |
---|
| 668 | |
---|
| 669 | SUBROUTINE transpose_zx( f_inv, f_out ) |
---|
| 670 | |
---|
| 671 | !------------------------------------------------------------------------------! |
---|
| 672 | ! Description: |
---|
| 673 | ! ------------ |
---|
[1] | 674 | ! Transposition of input array (f_in) from z to x. For the input array, all |
---|
| 675 | ! elements along z reside on the same PE, while after transposition, all |
---|
| 676 | ! elements along x reside on the same PE. |
---|
| 677 | !------------------------------------------------------------------------------! |
---|
| 678 | |
---|
| 679 | USE cpulog |
---|
| 680 | USE indices |
---|
| 681 | USE interfaces |
---|
| 682 | USE pegrid |
---|
| 683 | USE transpose_indices |
---|
| 684 | |
---|
| 685 | IMPLICIT NONE |
---|
| 686 | |
---|
[1111] | 687 | INTEGER :: i, j, k, l, xs |
---|
[1] | 688 | |
---|
[1216] | 689 | REAL :: f_inv(nys:nyn,nxl:nxr,1:nz), f_out(0:nx,nys_x:nyn_x,nzb_x:nzt_x) |
---|
[1] | 690 | |
---|
[1111] | 691 | REAL, DIMENSION(nys_x:nyn_x,nnx,nzb_x:nzt_x,0:pdims(1)-1) :: work |
---|
| 692 | |
---|
[1] | 693 | |
---|
| 694 | ! |
---|
| 695 | !-- If the PE grid is one-dimensional along y, only local reordering |
---|
| 696 | !-- of the data is necessary and no transposition has to be done. |
---|
| 697 | IF ( pdims(1) == 1 ) THEN |
---|
[1106] | 698 | |
---|
[683] | 699 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 700 | !$OMP DO |
---|
[1216] | 701 | !$acc kernels present( f_inv, f_out ) |
---|
[1003] | 702 | DO k = 1, nz |
---|
| 703 | DO i = nxl, nxr |
---|
| 704 | DO j = nys, nyn |
---|
[164] | 705 | f_out(i,j,k) = f_inv(j,i,k) |
---|
[1] | 706 | ENDDO |
---|
| 707 | ENDDO |
---|
| 708 | ENDDO |
---|
[1111] | 709 | !$acc end kernels |
---|
[683] | 710 | !$OMP END PARALLEL |
---|
[1] | 711 | |
---|
[1106] | 712 | ELSE |
---|
| 713 | |
---|
| 714 | #if defined( __parallel ) |
---|
[1] | 715 | ! |
---|
[1106] | 716 | !-- Transpose array |
---|
| 717 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
| 718 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1111] | 719 | !$acc update host( f_inv ) |
---|
| 720 | CALL MPI_ALLTOALL( f_inv(nys,nxl,1), sendrecvcount_zx, MPI_REAL, & |
---|
| 721 | work(nys_x,1,nzb_x,0), sendrecvcount_zx, MPI_REAL, & |
---|
[1106] | 722 | comm1dx, ierr ) |
---|
| 723 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
[1] | 724 | |
---|
| 725 | ! |
---|
[1106] | 726 | !-- Reorder transposed array |
---|
[1111] | 727 | !$OMP PARALLEL PRIVATE ( i, j, k, l, xs ) |
---|
[683] | 728 | !$OMP DO |
---|
[1216] | 729 | !$acc data copyin( work ) |
---|
[1106] | 730 | DO l = 0, pdims(1) - 1 |
---|
| 731 | xs = 0 + l * nnx |
---|
[1216] | 732 | !$acc kernels present( f_out ) |
---|
[1106] | 733 | DO k = nzb_x, nzt_x |
---|
| 734 | DO i = xs, xs + nnx - 1 |
---|
| 735 | DO j = nys_x, nyn_x |
---|
[1111] | 736 | f_out(i,j,k) = work(j,i-xs+1,k,l) |
---|
[1106] | 737 | ENDDO |
---|
[1] | 738 | ENDDO |
---|
| 739 | ENDDO |
---|
[1111] | 740 | !$acc end kernels |
---|
[1] | 741 | ENDDO |
---|
[1216] | 742 | !$acc end data |
---|
[683] | 743 | !$OMP END PARALLEL |
---|
[1] | 744 | #endif |
---|
| 745 | |
---|
[1106] | 746 | ENDIF |
---|
| 747 | |
---|
[1] | 748 | END SUBROUTINE transpose_zx |
---|
| 749 | |
---|
| 750 | |
---|
[1216] | 751 | SUBROUTINE resort_for_zy( f_inv, f_out ) |
---|
[1] | 752 | |
---|
| 753 | !------------------------------------------------------------------------------! |
---|
| 754 | ! Description: |
---|
| 755 | ! ------------ |
---|
[1216] | 756 | ! Resorting data after the transposition from z to y. The transposition itself |
---|
| 757 | ! is carried out in transpose_zy |
---|
| 758 | !------------------------------------------------------------------------------! |
---|
| 759 | |
---|
| 760 | USE indices |
---|
| 761 | USE transpose_indices |
---|
| 762 | |
---|
| 763 | IMPLICIT NONE |
---|
| 764 | |
---|
| 765 | REAL :: f_inv(nxl_y:nxr_y,nzb_y:nzt_y,0:ny) |
---|
| 766 | REAL :: f_out(0:ny,nxl_y:nxr_y,nzb_y:nzt_y) |
---|
| 767 | |
---|
| 768 | |
---|
| 769 | INTEGER :: i, j, k |
---|
| 770 | |
---|
| 771 | ! |
---|
| 772 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 773 | !-- by MPI contiguous |
---|
| 774 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 775 | !$OMP DO |
---|
| 776 | !$acc kernels present( f_inv, f_out ) |
---|
| 777 | DO k = nzb_y, nzt_y |
---|
| 778 | DO j = 0, ny |
---|
| 779 | DO i = nxl_y, nxr_y |
---|
| 780 | f_out(j,i,k) = f_inv(i,k,j) |
---|
| 781 | ENDDO |
---|
| 782 | ENDDO |
---|
| 783 | ENDDO |
---|
| 784 | !$acc end kernels |
---|
| 785 | !$OMP END PARALLEL |
---|
| 786 | |
---|
| 787 | END SUBROUTINE resort_for_zy |
---|
| 788 | |
---|
| 789 | |
---|
| 790 | SUBROUTINE transpose_zy( f_in, f_inv ) |
---|
| 791 | |
---|
| 792 | !------------------------------------------------------------------------------! |
---|
| 793 | ! Description: |
---|
| 794 | ! ------------ |
---|
[1] | 795 | ! Transposition of input array (f_in) from z to y. For the input array, all |
---|
| 796 | ! elements along z reside on the same PE, while after transposition, all |
---|
| 797 | ! elements along y reside on the same PE. |
---|
| 798 | !------------------------------------------------------------------------------! |
---|
| 799 | |
---|
| 800 | USE cpulog |
---|
| 801 | USE indices |
---|
| 802 | USE interfaces |
---|
| 803 | USE pegrid |
---|
| 804 | USE transpose_indices |
---|
| 805 | |
---|
| 806 | IMPLICIT NONE |
---|
| 807 | |
---|
[1111] | 808 | INTEGER :: i, j, k, l, zs |
---|
[1] | 809 | |
---|
[1216] | 810 | REAL :: f_in(nxl_z:nxr_z,nys_z:nyn_z,1:nz), f_inv(nxl_y:nxr_y,nzb_y:nzt_y,0:ny) |
---|
[1] | 811 | |
---|
[1111] | 812 | REAL, DIMENSION(nxl_z:nxr_z,nzt_y-nzb_y+1,nys_z:nyn_z,0:pdims(1)-1) :: work |
---|
| 813 | |
---|
| 814 | |
---|
[1] | 815 | ! |
---|
| 816 | !-- If the PE grid is one-dimensional along y, the array has only to be |
---|
| 817 | !-- reordered locally and therefore no transposition has to be done. |
---|
| 818 | IF ( pdims(1) /= 1 ) THEN |
---|
[1106] | 819 | |
---|
| 820 | #if defined( __parallel ) |
---|
[1] | 821 | ! |
---|
| 822 | !-- Reorder input array for transposition |
---|
[1111] | 823 | !$OMP PARALLEL PRIVATE ( i, j, k, l, zs ) |
---|
[683] | 824 | !$OMP DO |
---|
[1216] | 825 | !$acc data copyout( work ) |
---|
[1] | 826 | DO l = 0, pdims(1) - 1 |
---|
[1003] | 827 | zs = 1 + l * ( nzt_y - nzb_y + 1 ) |
---|
[1111] | 828 | !$acc kernels present( f_in, work ) |
---|
[1003] | 829 | DO j = nys_z, nyn_z |
---|
| 830 | DO k = zs, zs + nzt_y - nzb_y |
---|
| 831 | DO i = nxl_z, nxr_z |
---|
[1111] | 832 | work(i,k-zs+1,j,l) = f_in(i,j,k) |
---|
[1] | 833 | ENDDO |
---|
| 834 | ENDDO |
---|
| 835 | ENDDO |
---|
[1111] | 836 | !$acc end kernels |
---|
[1] | 837 | ENDDO |
---|
[1216] | 838 | !$acc end data |
---|
[683] | 839 | !$OMP END PARALLEL |
---|
[1] | 840 | |
---|
| 841 | ! |
---|
| 842 | !-- Transpose array |
---|
| 843 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
[622] | 844 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1111] | 845 | CALL MPI_ALLTOALL( work(nxl_z,1,nys_z,0), sendrecvcount_yz, MPI_REAL, & |
---|
| 846 | f_inv(nxl_y,nzb_y,0), sendrecvcount_yz, MPI_REAL, & |
---|
[1] | 847 | comm1dx, ierr ) |
---|
[1111] | 848 | !$acc update device( f_inv ) |
---|
[1] | 849 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
[1106] | 850 | #endif |
---|
[1] | 851 | |
---|
| 852 | ELSE |
---|
| 853 | ! |
---|
[1106] | 854 | !-- Reorder the array in the same way like ALLTOALL did it |
---|
[683] | 855 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
---|
| 856 | !$OMP DO |
---|
[1216] | 857 | !$acc kernels present( f_in, f_inv ) |
---|
[1003] | 858 | DO k = nzb_y, nzt_y |
---|
| 859 | DO j = 0, ny |
---|
| 860 | DO i = nxl_y, nxr_y |
---|
[164] | 861 | f_inv(i,k,j) = f_in(i,j,k) |
---|
| 862 | ENDDO |
---|
| 863 | ENDDO |
---|
| 864 | ENDDO |
---|
[1111] | 865 | !$acc end kernels |
---|
[683] | 866 | !$OMP END PARALLEL |
---|
[1106] | 867 | |
---|
| 868 | ENDIF |
---|
| 869 | |
---|
[1] | 870 | END SUBROUTINE transpose_zy |
---|
| 871 | |
---|
| 872 | |
---|
[1216] | 873 | SUBROUTINE transpose_zyd( f_in, f_out ) |
---|
[1] | 874 | |
---|
| 875 | !------------------------------------------------------------------------------! |
---|
| 876 | ! Description: |
---|
| 877 | ! ------------ |
---|
| 878 | ! Transposition of input array (f_in) from z to y. For the input array, all |
---|
| 879 | ! elements along z reside on the same PE, while after transposition, all |
---|
| 880 | ! elements along y reside on the same PE. |
---|
| 881 | ! This is a direct transposition for arrays with indices in regular order |
---|
| 882 | ! (k,j,i) (cf. transpose_zy). |
---|
| 883 | !------------------------------------------------------------------------------! |
---|
| 884 | |
---|
| 885 | USE cpulog |
---|
| 886 | USE indices |
---|
| 887 | USE interfaces |
---|
| 888 | USE pegrid |
---|
| 889 | USE transpose_indices |
---|
| 890 | |
---|
| 891 | IMPLICIT NONE |
---|
| 892 | |
---|
| 893 | INTEGER :: i, j, k, l, m, ys |
---|
| 894 | |
---|
[1003] | 895 | REAL :: f_in(1:nz,nys:nyn,nxl:nxr), f_inv(nys:nyn,nxl:nxr,1:nz), & |
---|
| 896 | f_out(0:ny,nxl_yd:nxr_yd,nzb_yd:nzt_yd), & |
---|
[164] | 897 | work(nnx*nny*nnz) |
---|
[1] | 898 | |
---|
| 899 | #if defined( __parallel ) |
---|
| 900 | |
---|
| 901 | ! |
---|
| 902 | !-- Rearrange indices of input array in order to make data to be send |
---|
| 903 | !-- by MPI contiguous |
---|
[1003] | 904 | DO i = nxl, nxr |
---|
| 905 | DO j = nys, nyn |
---|
| 906 | DO k = 1, nz |
---|
[164] | 907 | f_inv(j,i,k) = f_in(k,j,i) |
---|
[1] | 908 | ENDDO |
---|
| 909 | ENDDO |
---|
| 910 | ENDDO |
---|
| 911 | |
---|
| 912 | ! |
---|
| 913 | !-- Move data to different array, because memory location of work1 is |
---|
| 914 | !-- needed further below (work1 = work2). |
---|
| 915 | !-- If the PE grid is one-dimensional along x, only local reordering |
---|
| 916 | !-- of the data is necessary and no transposition has to be done. |
---|
| 917 | IF ( pdims(2) == 1 ) THEN |
---|
[1003] | 918 | DO k = 1, nz |
---|
| 919 | DO i = nxl, nxr |
---|
| 920 | DO j = nys, nyn |
---|
[164] | 921 | f_out(j,i,k) = f_inv(j,i,k) |
---|
[1] | 922 | ENDDO |
---|
| 923 | ENDDO |
---|
| 924 | ENDDO |
---|
| 925 | RETURN |
---|
| 926 | ENDIF |
---|
| 927 | |
---|
| 928 | ! |
---|
| 929 | !-- Transpose array |
---|
| 930 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
---|
[622] | 931 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1] | 932 | CALL MPI_ALLTOALL( f_inv(nys,nxl,1), sendrecvcount_zyd, MPI_REAL, & |
---|
[164] | 933 | work(1), sendrecvcount_zyd, MPI_REAL, & |
---|
[1] | 934 | comm1dy, ierr ) |
---|
| 935 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
---|
| 936 | |
---|
| 937 | ! |
---|
| 938 | !-- Reorder transposed array |
---|
| 939 | m = 0 |
---|
| 940 | DO l = 0, pdims(2) - 1 |
---|
| 941 | ys = 0 + l * nny |
---|
[1003] | 942 | DO k = nzb_yd, nzt_yd |
---|
| 943 | DO i = nxl_yd, nxr_yd |
---|
[1] | 944 | DO j = ys, ys + nny - 1 |
---|
| 945 | m = m + 1 |
---|
[164] | 946 | f_out(j,i,k) = work(m) |
---|
[1] | 947 | ENDDO |
---|
| 948 | ENDDO |
---|
| 949 | ENDDO |
---|
| 950 | ENDDO |
---|
| 951 | |
---|
| 952 | #endif |
---|
| 953 | |
---|
| 954 | END SUBROUTINE transpose_zyd |
---|