| 1 | SUBROUTINE transpose_xy( f_in, work, f_out ) |
|---|
| 2 | |
|---|
| 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 | ! |
|---|
| 20 | ! Current revisions: |
|---|
| 21 | ! ----------------- |
|---|
| 22 | ! |
|---|
| 23 | ! |
|---|
| 24 | ! Former revisions: |
|---|
| 25 | ! ----------------- |
|---|
| 26 | ! $Id: transpose.f90 1112 2013-03-09 00:34:37Z hoffmann $ |
|---|
| 27 | ! |
|---|
| 28 | ! 1111 2013-03-08 23:54:10Z raasch |
|---|
| 29 | ! openACC directives added, |
|---|
| 30 | ! resorting data from/to work changed, work got 4 dimensions instead of 1 |
|---|
| 31 | ! |
|---|
| 32 | ! 1106 2013-03-04 05:31:38Z raasch |
|---|
| 33 | ! preprocessor lines rearranged so that routines can also be used in serial |
|---|
| 34 | ! (non-parallel) mode |
|---|
| 35 | ! |
|---|
| 36 | ! 1092 2013-02-02 11:24:22Z raasch |
|---|
| 37 | ! unused variables removed |
|---|
| 38 | ! |
|---|
| 39 | ! 1036 2012-10-22 13:43:42Z raasch |
|---|
| 40 | ! code put under GPL (PALM 3.9) |
|---|
| 41 | ! |
|---|
| 42 | ! 1003 2012-09-14 14:35:53Z raasch |
|---|
| 43 | ! indices nxa, nya, etc. replaced by nx, ny, etc. |
|---|
| 44 | ! |
|---|
| 45 | ! 683 2011-02-09 14:25:15Z raasch |
|---|
| 46 | ! openMP parallelization of transpositions for 2d-domain-decomposition |
|---|
| 47 | ! |
|---|
| 48 | ! 622 2010-12-10 08:08:13Z raasch |
|---|
| 49 | ! optional barriers included in order to speed up collective operations |
|---|
| 50 | ! |
|---|
| 51 | ! 164 2008-05-15 08:46:15Z raasch |
|---|
| 52 | ! f_inv changed from subroutine argument to automatic array in order to do |
|---|
| 53 | ! re-ordering from f_in to f_inv in one step, one array work is needed instead |
|---|
| 54 | ! of work1 and work2 |
|---|
| 55 | ! |
|---|
| 56 | ! February 2007 |
|---|
| 57 | ! RCS Log replace by Id keyword, revision history cleaned up |
|---|
| 58 | ! |
|---|
| 59 | ! Revision 1.2 2004/04/30 13:12:17 raasch |
|---|
| 60 | ! Switched from mpi_alltoallv to the simpler mpi_alltoall, |
|---|
| 61 | ! all former transpose-routine files collected in this file, enlarged |
|---|
| 62 | ! transposition arrays introduced |
|---|
| 63 | ! |
|---|
| 64 | ! Revision 1.1 2004/04/30 13:08:16 raasch |
|---|
| 65 | ! Initial revision (collection of former routines transpose_xy, transpose_xz, |
|---|
| 66 | ! transpose_yx, transpose_yz, transpose_zx, transpose_zy) |
|---|
| 67 | ! |
|---|
| 68 | ! Revision 1.1 1997/07/24 11:25:18 raasch |
|---|
| 69 | ! Initial revision |
|---|
| 70 | ! |
|---|
| 71 | ! |
|---|
| 72 | ! Description: |
|---|
| 73 | ! ------------ |
|---|
| 74 | ! Transposition of input array (f_in) from x to y. For the input array, all |
|---|
| 75 | ! elements along x reside on the same PE, while after transposition, all |
|---|
| 76 | ! elements along y reside on the same PE. |
|---|
| 77 | !------------------------------------------------------------------------------! |
|---|
| 78 | |
|---|
| 79 | USE cpulog |
|---|
| 80 | USE indices |
|---|
| 81 | USE interfaces |
|---|
| 82 | USE pegrid |
|---|
| 83 | USE transpose_indices |
|---|
| 84 | |
|---|
| 85 | IMPLICIT NONE |
|---|
| 86 | |
|---|
| 87 | INTEGER :: i, j, k, l, ys |
|---|
| 88 | |
|---|
| 89 | REAL :: f_in(0:nx,nys_x:nyn_x,nzb_x:nzt_x), f_out(0:ny,nxl_y:nxr_y,nzb_y:nzt_y) |
|---|
| 90 | |
|---|
| 91 | REAL, DIMENSION(nyn_x-nys_x+1,nzb_y:nzt_y,nxl_y:nxr_y,0:pdims(2)-1) :: work |
|---|
| 92 | |
|---|
| 93 | !$acc declare create( f_inv ) |
|---|
| 94 | REAL :: f_inv(nys_x:nyn_x,nzb_x:nzt_x,0:nx) |
|---|
| 95 | |
|---|
| 96 | |
|---|
| 97 | ! |
|---|
| 98 | !-- Rearrange indices of input array in order to make data to be send |
|---|
| 99 | !-- by MPI contiguous |
|---|
| 100 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 101 | !$OMP DO |
|---|
| 102 | !$acc kernels present( f_in ) |
|---|
| 103 | !$acc loop |
|---|
| 104 | DO i = 0, nx |
|---|
| 105 | DO k = nzb_x, nzt_x |
|---|
| 106 | !$acc loop vector( 32 ) |
|---|
| 107 | DO j = nys_x, nyn_x |
|---|
| 108 | f_inv(j,k,i) = f_in(i,j,k) |
|---|
| 109 | ENDDO |
|---|
| 110 | ENDDO |
|---|
| 111 | ENDDO |
|---|
| 112 | !$acc end kernels |
|---|
| 113 | !$OMP END PARALLEL |
|---|
| 114 | |
|---|
| 115 | IF ( numprocs /= 1 ) THEN |
|---|
| 116 | |
|---|
| 117 | #if defined( __parallel ) |
|---|
| 118 | ! |
|---|
| 119 | !-- Transpose array |
|---|
| 120 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 121 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 122 | !$acc update host( f_inv ) |
|---|
| 123 | CALL MPI_ALLTOALL( f_inv(nys_x,nzb_x,0), sendrecvcount_xy, MPI_REAL, & |
|---|
| 124 | work(1,nzb_y,nxl_y,0), sendrecvcount_xy, MPI_REAL, & |
|---|
| 125 | comm1dy, ierr ) |
|---|
| 126 | !$acc update device( work ) |
|---|
| 127 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 128 | |
|---|
| 129 | ! |
|---|
| 130 | !-- Reorder transposed array |
|---|
| 131 | !$OMP PARALLEL PRIVATE ( i, j, k, l, ys ) |
|---|
| 132 | !$OMP DO |
|---|
| 133 | DO l = 0, pdims(2) - 1 |
|---|
| 134 | ys = 0 + l * ( nyn_x - nys_x + 1 ) |
|---|
| 135 | !$acc kernels present( f_out, work ) |
|---|
| 136 | !$acc loop |
|---|
| 137 | DO i = nxl_y, nxr_y |
|---|
| 138 | DO k = nzb_y, nzt_y |
|---|
| 139 | !$acc loop vector( 32 ) |
|---|
| 140 | DO j = ys, ys + nyn_x - nys_x |
|---|
| 141 | f_out(j,i,k) = work(j-ys+1,k,i,l) |
|---|
| 142 | ENDDO |
|---|
| 143 | ENDDO |
|---|
| 144 | ENDDO |
|---|
| 145 | !$acc end kernels |
|---|
| 146 | ENDDO |
|---|
| 147 | !$OMP END PARALLEL |
|---|
| 148 | #endif |
|---|
| 149 | |
|---|
| 150 | ELSE |
|---|
| 151 | |
|---|
| 152 | ! |
|---|
| 153 | !-- Reorder transposed array |
|---|
| 154 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 155 | !$OMP DO |
|---|
| 156 | !$acc kernels present( f_out ) |
|---|
| 157 | !$acc loop |
|---|
| 158 | DO k = nzb_y, nzt_y |
|---|
| 159 | DO i = nxl_y, nxr_y |
|---|
| 160 | !$acc loop vector( 32 ) |
|---|
| 161 | DO j = 0, ny |
|---|
| 162 | f_out(j,i,k) = f_inv(j,k,i) |
|---|
| 163 | ENDDO |
|---|
| 164 | ENDDO |
|---|
| 165 | ENDDO |
|---|
| 166 | !$acc end kernels |
|---|
| 167 | !$OMP END PARALLEL |
|---|
| 168 | |
|---|
| 169 | ENDIF |
|---|
| 170 | |
|---|
| 171 | END SUBROUTINE transpose_xy |
|---|
| 172 | |
|---|
| 173 | |
|---|
| 174 | SUBROUTINE transpose_xz( f_in, work, f_out ) |
|---|
| 175 | |
|---|
| 176 | !------------------------------------------------------------------------------! |
|---|
| 177 | ! Description: |
|---|
| 178 | ! ------------ |
|---|
| 179 | ! Transposition of input array (f_in) from x to z. For the input array, all |
|---|
| 180 | ! elements along x reside on the same PE, while after transposition, all |
|---|
| 181 | ! elements along z reside on the same PE. |
|---|
| 182 | !------------------------------------------------------------------------------! |
|---|
| 183 | |
|---|
| 184 | USE cpulog |
|---|
| 185 | USE indices |
|---|
| 186 | USE interfaces |
|---|
| 187 | USE pegrid |
|---|
| 188 | USE transpose_indices |
|---|
| 189 | |
|---|
| 190 | IMPLICIT NONE |
|---|
| 191 | |
|---|
| 192 | INTEGER :: i, j, k, l, xs |
|---|
| 193 | |
|---|
| 194 | REAL :: f_in(0:nx,nys_x:nyn_x,nzb_x:nzt_x), f_out(1:nz,nys:nyn,nxl:nxr) |
|---|
| 195 | |
|---|
| 196 | REAL, DIMENSION(nys_x:nyn_x,nnx,nzb_x:nzt_x,0:pdims(1)-1) :: work |
|---|
| 197 | |
|---|
| 198 | !$acc declare create( f_inv ) |
|---|
| 199 | REAL :: f_inv(nys:nyn,nxl:nxr,1:nz) |
|---|
| 200 | |
|---|
| 201 | |
|---|
| 202 | ! |
|---|
| 203 | !-- If the PE grid is one-dimensional along y, the array has only to be |
|---|
| 204 | !-- reordered locally and therefore no transposition has to be done. |
|---|
| 205 | IF ( pdims(1) /= 1 ) THEN |
|---|
| 206 | |
|---|
| 207 | #if defined( __parallel ) |
|---|
| 208 | ! |
|---|
| 209 | !-- Reorder input array for transposition |
|---|
| 210 | !$OMP PARALLEL PRIVATE ( i, j, k, l, xs ) |
|---|
| 211 | !$OMP DO |
|---|
| 212 | DO l = 0, pdims(1) - 1 |
|---|
| 213 | xs = 0 + l * nnx |
|---|
| 214 | !$acc kernels present( f_in, work ) |
|---|
| 215 | !$acc loop |
|---|
| 216 | DO k = nzb_x, nzt_x |
|---|
| 217 | DO i = xs, xs + nnx - 1 |
|---|
| 218 | !$acc loop vector( 32 ) |
|---|
| 219 | DO j = nys_x, nyn_x |
|---|
| 220 | work(j,i-xs+1,k,l) = f_in(i,j,k) |
|---|
| 221 | ENDDO |
|---|
| 222 | ENDDO |
|---|
| 223 | ENDDO |
|---|
| 224 | !$acc end kernels |
|---|
| 225 | ENDDO |
|---|
| 226 | !$OMP END PARALLEL |
|---|
| 227 | |
|---|
| 228 | ! |
|---|
| 229 | !-- Transpose array |
|---|
| 230 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 231 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 232 | !$acc update host( work ) |
|---|
| 233 | CALL MPI_ALLTOALL( work(nys_x,1,nzb_x,0), sendrecvcount_zx, MPI_REAL, & |
|---|
| 234 | f_inv(nys,nxl,1), sendrecvcount_zx, MPI_REAL, & |
|---|
| 235 | comm1dx, ierr ) |
|---|
| 236 | !$acc update device( f_inv ) |
|---|
| 237 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 238 | |
|---|
| 239 | ! |
|---|
| 240 | !-- Reorder transposed array in a way that the z index is in first position |
|---|
| 241 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 242 | !$OMP DO |
|---|
| 243 | !$acc kernels present( f_out ) |
|---|
| 244 | !$acc loop |
|---|
| 245 | DO k = 1, nz |
|---|
| 246 | DO i = nxl, nxr |
|---|
| 247 | !$acc loop vector( 32 ) |
|---|
| 248 | DO j = nys, nyn |
|---|
| 249 | f_out(k,j,i) = f_inv(j,i,k) |
|---|
| 250 | ENDDO |
|---|
| 251 | ENDDO |
|---|
| 252 | ENDDO |
|---|
| 253 | !$acc end kernels |
|---|
| 254 | !$OMP END PARALLEL |
|---|
| 255 | #endif |
|---|
| 256 | |
|---|
| 257 | ELSE |
|---|
| 258 | |
|---|
| 259 | ! |
|---|
| 260 | !-- Reorder the array in a way that the z index is in first position |
|---|
| 261 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 262 | !$OMP DO |
|---|
| 263 | !$acc kernels present( f_in ) |
|---|
| 264 | !$acc loop |
|---|
| 265 | DO i = nxl, nxr |
|---|
| 266 | DO j = nys, nyn |
|---|
| 267 | !$acc loop vector( 32 ) |
|---|
| 268 | DO k = 1, nz |
|---|
| 269 | f_inv(j,i,k) = f_in(i,j,k) |
|---|
| 270 | ENDDO |
|---|
| 271 | ENDDO |
|---|
| 272 | ENDDO |
|---|
| 273 | !$acc end kernels |
|---|
| 274 | !$OMP END PARALLEL |
|---|
| 275 | |
|---|
| 276 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 277 | !$OMP DO |
|---|
| 278 | !$acc kernels present( f_out ) |
|---|
| 279 | !$acc loop |
|---|
| 280 | DO k = 1, nz |
|---|
| 281 | DO i = nxl, nxr |
|---|
| 282 | !$acc loop vector( 32 ) |
|---|
| 283 | DO j = nys, nyn |
|---|
| 284 | f_out(k,j,i) = f_inv(j,i,k) |
|---|
| 285 | ENDDO |
|---|
| 286 | ENDDO |
|---|
| 287 | ENDDO |
|---|
| 288 | !$acc end kernels |
|---|
| 289 | !$OMP END PARALLEL |
|---|
| 290 | |
|---|
| 291 | ENDIF |
|---|
| 292 | |
|---|
| 293 | END SUBROUTINE transpose_xz |
|---|
| 294 | |
|---|
| 295 | |
|---|
| 296 | SUBROUTINE transpose_yx( f_in, work, f_out ) |
|---|
| 297 | |
|---|
| 298 | !------------------------------------------------------------------------------! |
|---|
| 299 | ! Description: |
|---|
| 300 | ! ------------ |
|---|
| 301 | ! Transposition of input array (f_in) from y to x. For the input array, all |
|---|
| 302 | ! elements along y reside on the same PE, while after transposition, all |
|---|
| 303 | ! elements along x reside on the same PE. |
|---|
| 304 | !------------------------------------------------------------------------------! |
|---|
| 305 | |
|---|
| 306 | USE cpulog |
|---|
| 307 | USE indices |
|---|
| 308 | USE interfaces |
|---|
| 309 | USE pegrid |
|---|
| 310 | USE transpose_indices |
|---|
| 311 | |
|---|
| 312 | IMPLICIT NONE |
|---|
| 313 | |
|---|
| 314 | INTEGER :: i, j, k, l, ys |
|---|
| 315 | |
|---|
| 316 | REAL :: f_in(0:ny,nxl_y:nxr_y,nzb_y:nzt_y), f_out(0:nx,nys_x:nyn_x,nzb_x:nzt_x) |
|---|
| 317 | |
|---|
| 318 | REAL, DIMENSION(nyn_x-nys_x+1,nzb_y:nzt_y,nxl_y:nxr_y,0:pdims(2)-1) :: work |
|---|
| 319 | |
|---|
| 320 | !$acc declare create( f_inv ) |
|---|
| 321 | REAL :: f_inv(nys_x:nyn_x,nzb_x:nzt_x,0:nx) |
|---|
| 322 | |
|---|
| 323 | |
|---|
| 324 | IF ( numprocs /= 1 ) THEN |
|---|
| 325 | |
|---|
| 326 | #if defined( __parallel ) |
|---|
| 327 | ! |
|---|
| 328 | !-- Reorder input array for transposition |
|---|
| 329 | !$OMP PARALLEL PRIVATE ( i, j, k, l, ys ) |
|---|
| 330 | !$OMP DO |
|---|
| 331 | DO l = 0, pdims(2) - 1 |
|---|
| 332 | ys = 0 + l * ( nyn_x - nys_x + 1 ) |
|---|
| 333 | !$acc kernels present( f_in, work ) |
|---|
| 334 | !$acc loop |
|---|
| 335 | DO i = nxl_y, nxr_y |
|---|
| 336 | DO k = nzb_y, nzt_y |
|---|
| 337 | !$acc loop vector( 32 ) |
|---|
| 338 | DO j = ys, ys + nyn_x - nys_x |
|---|
| 339 | work(j-ys+1,k,i,l) = f_in(j,i,k) |
|---|
| 340 | ENDDO |
|---|
| 341 | ENDDO |
|---|
| 342 | ENDDO |
|---|
| 343 | !$acc end kernels |
|---|
| 344 | ENDDO |
|---|
| 345 | !$OMP END PARALLEL |
|---|
| 346 | |
|---|
| 347 | ! |
|---|
| 348 | !-- Transpose array |
|---|
| 349 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 350 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 351 | !$acc update host( work ) |
|---|
| 352 | CALL MPI_ALLTOALL( work(1,nzb_y,nxl_y,0), sendrecvcount_xy, MPI_REAL, & |
|---|
| 353 | f_inv(nys_x,nzb_x,0), sendrecvcount_xy, MPI_REAL, & |
|---|
| 354 | comm1dy, ierr ) |
|---|
| 355 | !$acc update device( f_inv ) |
|---|
| 356 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 357 | #endif |
|---|
| 358 | |
|---|
| 359 | ELSE |
|---|
| 360 | |
|---|
| 361 | ! |
|---|
| 362 | !-- Reorder array f_in the same way as ALLTOALL did it |
|---|
| 363 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 364 | !$OMP DO |
|---|
| 365 | !$acc kernels present( f_in ) |
|---|
| 366 | !$acc loop |
|---|
| 367 | DO i = nxl_y, nxr_y |
|---|
| 368 | DO k = nzb_y, nzt_y |
|---|
| 369 | !$acc loop vector( 32 ) |
|---|
| 370 | DO j = 0, ny |
|---|
| 371 | f_inv(j,k,i) = f_in(j,i,k) |
|---|
| 372 | ENDDO |
|---|
| 373 | ENDDO |
|---|
| 374 | ENDDO |
|---|
| 375 | !$acc end kernels |
|---|
| 376 | !$OMP END PARALLEL |
|---|
| 377 | |
|---|
| 378 | ENDIF |
|---|
| 379 | |
|---|
| 380 | ! |
|---|
| 381 | !-- Reorder transposed array in a way that the x index is in first position |
|---|
| 382 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 383 | !$OMP DO |
|---|
| 384 | !$acc kernels present( f_out ) |
|---|
| 385 | !$acc loop |
|---|
| 386 | DO i = 0, nx |
|---|
| 387 | DO k = nzb_x, nzt_x |
|---|
| 388 | !$acc loop vector( 32 ) |
|---|
| 389 | DO j = nys_x, nyn_x |
|---|
| 390 | f_out(i,j,k) = f_inv(j,k,i) |
|---|
| 391 | ENDDO |
|---|
| 392 | ENDDO |
|---|
| 393 | ENDDO |
|---|
| 394 | !$acc end kernels |
|---|
| 395 | !$OMP END PARALLEL |
|---|
| 396 | |
|---|
| 397 | END SUBROUTINE transpose_yx |
|---|
| 398 | |
|---|
| 399 | |
|---|
| 400 | SUBROUTINE transpose_yxd( f_in, work, f_out ) |
|---|
| 401 | |
|---|
| 402 | !------------------------------------------------------------------------------! |
|---|
| 403 | ! Description: |
|---|
| 404 | ! ------------ |
|---|
| 405 | ! Transposition of input array (f_in) from y to x. For the input array, all |
|---|
| 406 | ! elements along y reside on the same PE, while after transposition, all |
|---|
| 407 | ! elements along x reside on the same PE. |
|---|
| 408 | ! This is a direct transposition for arrays with indices in regular order |
|---|
| 409 | ! (k,j,i) (cf. transpose_yx). |
|---|
| 410 | !------------------------------------------------------------------------------! |
|---|
| 411 | |
|---|
| 412 | USE cpulog |
|---|
| 413 | USE indices |
|---|
| 414 | USE interfaces |
|---|
| 415 | USE pegrid |
|---|
| 416 | USE transpose_indices |
|---|
| 417 | |
|---|
| 418 | IMPLICIT NONE |
|---|
| 419 | |
|---|
| 420 | INTEGER :: i, j, k, l, m, xs |
|---|
| 421 | |
|---|
| 422 | REAL :: f_in(1:nz,nys:nyn,nxl:nxr), f_inv(nxl:nxr,1:nz,nys:nyn), & |
|---|
| 423 | f_out(0:nx,nys_x:nyn_x,nzb_x:nzt_x), & |
|---|
| 424 | work(nnx*nny*nnz) |
|---|
| 425 | |
|---|
| 426 | #if defined( __parallel ) |
|---|
| 427 | |
|---|
| 428 | ! |
|---|
| 429 | !-- Rearrange indices of input array in order to make data to be send |
|---|
| 430 | !-- by MPI contiguous |
|---|
| 431 | DO k = 1, nz |
|---|
| 432 | DO j = nys, nyn |
|---|
| 433 | DO i = nxl, nxr |
|---|
| 434 | f_inv(i,k,j) = f_in(k,j,i) |
|---|
| 435 | ENDDO |
|---|
| 436 | ENDDO |
|---|
| 437 | ENDDO |
|---|
| 438 | |
|---|
| 439 | ! |
|---|
| 440 | !-- Transpose array |
|---|
| 441 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 442 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 443 | CALL MPI_ALLTOALL( f_inv(nxl,1,nys), sendrecvcount_xy, MPI_REAL, & |
|---|
| 444 | work(1), sendrecvcount_xy, MPI_REAL, & |
|---|
| 445 | comm1dx, ierr ) |
|---|
| 446 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 447 | |
|---|
| 448 | ! |
|---|
| 449 | !-- Reorder transposed array |
|---|
| 450 | m = 0 |
|---|
| 451 | DO l = 0, pdims(1) - 1 |
|---|
| 452 | xs = 0 + l * nnx |
|---|
| 453 | DO j = nys_x, nyn_x |
|---|
| 454 | DO k = 1, nz |
|---|
| 455 | DO i = xs, xs + nnx - 1 |
|---|
| 456 | m = m + 1 |
|---|
| 457 | f_out(i,j,k) = work(m) |
|---|
| 458 | ENDDO |
|---|
| 459 | ENDDO |
|---|
| 460 | ENDDO |
|---|
| 461 | ENDDO |
|---|
| 462 | |
|---|
| 463 | #endif |
|---|
| 464 | |
|---|
| 465 | END SUBROUTINE transpose_yxd |
|---|
| 466 | |
|---|
| 467 | |
|---|
| 468 | SUBROUTINE transpose_yz( f_in, work, f_out ) |
|---|
| 469 | |
|---|
| 470 | !------------------------------------------------------------------------------! |
|---|
| 471 | ! Description: |
|---|
| 472 | ! ------------ |
|---|
| 473 | ! Transposition of input array (f_in) from y to z. For the input array, all |
|---|
| 474 | ! elements along y reside on the same PE, while after transposition, all |
|---|
| 475 | ! elements along z reside on the same PE. |
|---|
| 476 | !------------------------------------------------------------------------------! |
|---|
| 477 | |
|---|
| 478 | USE cpulog |
|---|
| 479 | USE indices |
|---|
| 480 | USE interfaces |
|---|
| 481 | USE pegrid |
|---|
| 482 | USE transpose_indices |
|---|
| 483 | |
|---|
| 484 | IMPLICIT NONE |
|---|
| 485 | |
|---|
| 486 | INTEGER :: i, j, k, l, zs |
|---|
| 487 | |
|---|
| 488 | REAL :: f_in(0:ny,nxl_y:nxr_y,nzb_y:nzt_y), f_out(nxl_z:nxr_z,nys_z:nyn_z,1:nz) |
|---|
| 489 | |
|---|
| 490 | REAL, DIMENSION(nxl_z:nxr_z,nzt_y-nzb_y+1,nys_z:nyn_z,0:pdims(1)-1) :: work |
|---|
| 491 | |
|---|
| 492 | !$acc declare create( f_inv ) |
|---|
| 493 | REAL :: f_inv(nxl_y:nxr_y,nzb_y:nzt_y,0:ny) |
|---|
| 494 | |
|---|
| 495 | |
|---|
| 496 | ! |
|---|
| 497 | !-- Rearrange indices of input array in order to make data to be send |
|---|
| 498 | !-- by MPI contiguous |
|---|
| 499 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 500 | !$OMP DO |
|---|
| 501 | !$acc kernels present( f_in ) |
|---|
| 502 | !$acc loop |
|---|
| 503 | DO j = 0, ny |
|---|
| 504 | DO k = nzb_y, nzt_y |
|---|
| 505 | !$acc loop vector( 32 ) |
|---|
| 506 | DO i = nxl_y, nxr_y |
|---|
| 507 | f_inv(i,k,j) = f_in(j,i,k) |
|---|
| 508 | ENDDO |
|---|
| 509 | ENDDO |
|---|
| 510 | ENDDO |
|---|
| 511 | !$acc end kernels |
|---|
| 512 | !$OMP END PARALLEL |
|---|
| 513 | |
|---|
| 514 | ! |
|---|
| 515 | !-- Move data to different array, because memory location of work1 is |
|---|
| 516 | !-- needed further below (work1 = work2). |
|---|
| 517 | !-- If the PE grid is one-dimensional along y, only local reordering |
|---|
| 518 | !-- of the data is necessary and no transposition has to be done. |
|---|
| 519 | IF ( pdims(1) == 1 ) THEN |
|---|
| 520 | |
|---|
| 521 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 522 | !$OMP DO |
|---|
| 523 | !$acc kernels present( f_out ) |
|---|
| 524 | !$acc loop |
|---|
| 525 | DO j = 0, ny |
|---|
| 526 | DO k = nzb_y, nzt_y |
|---|
| 527 | !$acc loop vector( 32 ) |
|---|
| 528 | DO i = nxl_y, nxr_y |
|---|
| 529 | f_out(i,j,k) = f_inv(i,k,j) |
|---|
| 530 | ENDDO |
|---|
| 531 | ENDDO |
|---|
| 532 | ENDDO |
|---|
| 533 | !$acc end kernels |
|---|
| 534 | !$OMP END PARALLEL |
|---|
| 535 | |
|---|
| 536 | ELSE |
|---|
| 537 | |
|---|
| 538 | #if defined( __parallel ) |
|---|
| 539 | ! |
|---|
| 540 | !-- Transpose array |
|---|
| 541 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 542 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 543 | !$acc update host( f_inv ) |
|---|
| 544 | CALL MPI_ALLTOALL( f_inv(nxl_y,nzb_y,0), sendrecvcount_yz, MPI_REAL, & |
|---|
| 545 | work(nxl_z,1,nys_z,0), sendrecvcount_yz, MPI_REAL, & |
|---|
| 546 | comm1dx, ierr ) |
|---|
| 547 | !$acc update device( work ) |
|---|
| 548 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 549 | |
|---|
| 550 | ! |
|---|
| 551 | !-- Reorder transposed array |
|---|
| 552 | !$OMP PARALLEL PRIVATE ( i, j, k, l, zs ) |
|---|
| 553 | !$OMP DO |
|---|
| 554 | DO l = 0, pdims(1) - 1 |
|---|
| 555 | zs = 1 + l * ( nzt_y - nzb_y + 1 ) |
|---|
| 556 | !$acc kernels present( f_out, work ) |
|---|
| 557 | !$acc loop |
|---|
| 558 | DO j = nys_z, nyn_z |
|---|
| 559 | DO k = zs, zs + nzt_y - nzb_y |
|---|
| 560 | !$acc loop vector( 32 ) |
|---|
| 561 | DO i = nxl_z, nxr_z |
|---|
| 562 | f_out(i,j,k) = work(i,k-zs+1,j,l) |
|---|
| 563 | ENDDO |
|---|
| 564 | ENDDO |
|---|
| 565 | ENDDO |
|---|
| 566 | !$acc end kernels |
|---|
| 567 | ENDDO |
|---|
| 568 | !$OMP END PARALLEL |
|---|
| 569 | #endif |
|---|
| 570 | |
|---|
| 571 | ENDIF |
|---|
| 572 | |
|---|
| 573 | END SUBROUTINE transpose_yz |
|---|
| 574 | |
|---|
| 575 | |
|---|
| 576 | SUBROUTINE transpose_zx( f_in, work, f_out ) |
|---|
| 577 | |
|---|
| 578 | !------------------------------------------------------------------------------! |
|---|
| 579 | ! Description: |
|---|
| 580 | ! ------------ |
|---|
| 581 | ! Transposition of input array (f_in) from z to x. For the input array, all |
|---|
| 582 | ! elements along z reside on the same PE, while after transposition, all |
|---|
| 583 | ! elements along x reside on the same PE. |
|---|
| 584 | !------------------------------------------------------------------------------! |
|---|
| 585 | |
|---|
| 586 | USE cpulog |
|---|
| 587 | USE indices |
|---|
| 588 | USE interfaces |
|---|
| 589 | USE pegrid |
|---|
| 590 | USE transpose_indices |
|---|
| 591 | |
|---|
| 592 | IMPLICIT NONE |
|---|
| 593 | |
|---|
| 594 | INTEGER :: i, j, k, l, xs |
|---|
| 595 | |
|---|
| 596 | REAL :: f_in(1:nz,nys:nyn,nxl:nxr), f_out(0:nx,nys_x:nyn_x,nzb_x:nzt_x) |
|---|
| 597 | |
|---|
| 598 | REAL, DIMENSION(nys_x:nyn_x,nnx,nzb_x:nzt_x,0:pdims(1)-1) :: work |
|---|
| 599 | |
|---|
| 600 | !$acc declare create( f_inv ) |
|---|
| 601 | REAL :: f_inv(nys:nyn,nxl:nxr,1:nz) |
|---|
| 602 | |
|---|
| 603 | |
|---|
| 604 | ! |
|---|
| 605 | !-- Rearrange indices of input array in order to make data to be send |
|---|
| 606 | !-- by MPI contiguous |
|---|
| 607 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 608 | !$OMP DO |
|---|
| 609 | !$acc kernels present( f_in ) |
|---|
| 610 | !$acc loop |
|---|
| 611 | DO k = 1,nz |
|---|
| 612 | DO i = nxl, nxr |
|---|
| 613 | !$acc loop vector( 32 ) |
|---|
| 614 | DO j = nys, nyn |
|---|
| 615 | f_inv(j,i,k) = f_in(k,j,i) |
|---|
| 616 | ENDDO |
|---|
| 617 | ENDDO |
|---|
| 618 | ENDDO |
|---|
| 619 | !$acc end kernels |
|---|
| 620 | !$OMP END PARALLEL |
|---|
| 621 | |
|---|
| 622 | ! |
|---|
| 623 | !-- Move data to different array, because memory location of work1 is |
|---|
| 624 | !-- needed further below (work1 = work2). |
|---|
| 625 | !-- If the PE grid is one-dimensional along y, only local reordering |
|---|
| 626 | !-- of the data is necessary and no transposition has to be done. |
|---|
| 627 | IF ( pdims(1) == 1 ) THEN |
|---|
| 628 | |
|---|
| 629 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 630 | !$OMP DO |
|---|
| 631 | !$acc kernels present( f_out ) |
|---|
| 632 | !$acc loop |
|---|
| 633 | DO k = 1, nz |
|---|
| 634 | DO i = nxl, nxr |
|---|
| 635 | !$acc loop vector( 32 ) |
|---|
| 636 | DO j = nys, nyn |
|---|
| 637 | f_out(i,j,k) = f_inv(j,i,k) |
|---|
| 638 | ENDDO |
|---|
| 639 | ENDDO |
|---|
| 640 | ENDDO |
|---|
| 641 | !$acc end kernels |
|---|
| 642 | !$OMP END PARALLEL |
|---|
| 643 | |
|---|
| 644 | ELSE |
|---|
| 645 | |
|---|
| 646 | #if defined( __parallel ) |
|---|
| 647 | ! |
|---|
| 648 | !-- Transpose array |
|---|
| 649 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 650 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 651 | !$acc update host( f_inv ) |
|---|
| 652 | CALL MPI_ALLTOALL( f_inv(nys,nxl,1), sendrecvcount_zx, MPI_REAL, & |
|---|
| 653 | work(nys_x,1,nzb_x,0), sendrecvcount_zx, MPI_REAL, & |
|---|
| 654 | comm1dx, ierr ) |
|---|
| 655 | !$acc update device( work ) |
|---|
| 656 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 657 | |
|---|
| 658 | ! |
|---|
| 659 | !-- Reorder transposed array |
|---|
| 660 | !$OMP PARALLEL PRIVATE ( i, j, k, l, xs ) |
|---|
| 661 | !$OMP DO |
|---|
| 662 | DO l = 0, pdims(1) - 1 |
|---|
| 663 | xs = 0 + l * nnx |
|---|
| 664 | !$acc kernels present( f_out, work ) |
|---|
| 665 | !$acc loop |
|---|
| 666 | DO k = nzb_x, nzt_x |
|---|
| 667 | DO i = xs, xs + nnx - 1 |
|---|
| 668 | !$acc loop vector( 32 ) |
|---|
| 669 | DO j = nys_x, nyn_x |
|---|
| 670 | f_out(i,j,k) = work(j,i-xs+1,k,l) |
|---|
| 671 | ENDDO |
|---|
| 672 | ENDDO |
|---|
| 673 | ENDDO |
|---|
| 674 | !$acc end kernels |
|---|
| 675 | ENDDO |
|---|
| 676 | !$OMP END PARALLEL |
|---|
| 677 | #endif |
|---|
| 678 | |
|---|
| 679 | ENDIF |
|---|
| 680 | |
|---|
| 681 | END SUBROUTINE transpose_zx |
|---|
| 682 | |
|---|
| 683 | |
|---|
| 684 | SUBROUTINE transpose_zy( f_in, work, f_out ) |
|---|
| 685 | |
|---|
| 686 | !------------------------------------------------------------------------------! |
|---|
| 687 | ! Description: |
|---|
| 688 | ! ------------ |
|---|
| 689 | ! Transposition of input array (f_in) from z to y. For the input array, all |
|---|
| 690 | ! elements along z reside on the same PE, while after transposition, all |
|---|
| 691 | ! elements along y reside on the same PE. |
|---|
| 692 | !------------------------------------------------------------------------------! |
|---|
| 693 | |
|---|
| 694 | USE cpulog |
|---|
| 695 | USE indices |
|---|
| 696 | USE interfaces |
|---|
| 697 | USE pegrid |
|---|
| 698 | USE transpose_indices |
|---|
| 699 | |
|---|
| 700 | IMPLICIT NONE |
|---|
| 701 | |
|---|
| 702 | INTEGER :: i, j, k, l, zs |
|---|
| 703 | |
|---|
| 704 | REAL :: f_in(nxl_z:nxr_z,nys_z:nyn_z,1:nz), f_out(0:ny,nxl_y:nxr_y,nzb_y:nzt_y) |
|---|
| 705 | |
|---|
| 706 | REAL, DIMENSION(nxl_z:nxr_z,nzt_y-nzb_y+1,nys_z:nyn_z,0:pdims(1)-1) :: work |
|---|
| 707 | |
|---|
| 708 | !$acc declare create( f_inv ) |
|---|
| 709 | REAL :: f_inv(nxl_y:nxr_y,nzb_y:nzt_y,0:ny) |
|---|
| 710 | |
|---|
| 711 | |
|---|
| 712 | ! |
|---|
| 713 | !-- If the PE grid is one-dimensional along y, the array has only to be |
|---|
| 714 | !-- reordered locally and therefore no transposition has to be done. |
|---|
| 715 | IF ( pdims(1) /= 1 ) THEN |
|---|
| 716 | |
|---|
| 717 | #if defined( __parallel ) |
|---|
| 718 | ! |
|---|
| 719 | !-- Reorder input array for transposition |
|---|
| 720 | !$OMP PARALLEL PRIVATE ( i, j, k, l, zs ) |
|---|
| 721 | !$OMP DO |
|---|
| 722 | DO l = 0, pdims(1) - 1 |
|---|
| 723 | zs = 1 + l * ( nzt_y - nzb_y + 1 ) |
|---|
| 724 | !$acc kernels present( f_in, work ) |
|---|
| 725 | !$acc loop |
|---|
| 726 | DO j = nys_z, nyn_z |
|---|
| 727 | DO k = zs, zs + nzt_y - nzb_y |
|---|
| 728 | !$acc loop vector( 32 ) |
|---|
| 729 | DO i = nxl_z, nxr_z |
|---|
| 730 | work(i,k-zs+1,j,l) = f_in(i,j,k) |
|---|
| 731 | ENDDO |
|---|
| 732 | ENDDO |
|---|
| 733 | ENDDO |
|---|
| 734 | !$acc end kernels |
|---|
| 735 | ENDDO |
|---|
| 736 | !$OMP END PARALLEL |
|---|
| 737 | |
|---|
| 738 | ! |
|---|
| 739 | !-- Transpose array |
|---|
| 740 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 741 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 742 | !$acc update host( work ) |
|---|
| 743 | CALL MPI_ALLTOALL( work(nxl_z,1,nys_z,0), sendrecvcount_yz, MPI_REAL, & |
|---|
| 744 | f_inv(nxl_y,nzb_y,0), sendrecvcount_yz, MPI_REAL, & |
|---|
| 745 | comm1dx, ierr ) |
|---|
| 746 | !$acc update device( f_inv ) |
|---|
| 747 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 748 | #endif |
|---|
| 749 | |
|---|
| 750 | ELSE |
|---|
| 751 | ! |
|---|
| 752 | !-- Reorder the array in the same way like ALLTOALL did it |
|---|
| 753 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 754 | !$OMP DO |
|---|
| 755 | !$acc kernels present( f_in ) |
|---|
| 756 | !$acc loop |
|---|
| 757 | DO k = nzb_y, nzt_y |
|---|
| 758 | DO j = 0, ny |
|---|
| 759 | !$acc loop vector( 32 ) |
|---|
| 760 | DO i = nxl_y, nxr_y |
|---|
| 761 | f_inv(i,k,j) = f_in(i,j,k) |
|---|
| 762 | ENDDO |
|---|
| 763 | ENDDO |
|---|
| 764 | ENDDO |
|---|
| 765 | !$acc end kernels |
|---|
| 766 | !$OMP END PARALLEL |
|---|
| 767 | |
|---|
| 768 | ENDIF |
|---|
| 769 | |
|---|
| 770 | ! |
|---|
| 771 | !-- Reorder transposed array in a way that the y index is in first position |
|---|
| 772 | !$OMP PARALLEL PRIVATE ( i, j, k ) |
|---|
| 773 | !$OMP DO |
|---|
| 774 | !$acc kernels present( f_out ) |
|---|
| 775 | !$acc loop |
|---|
| 776 | DO k = nzb_y, nzt_y |
|---|
| 777 | DO i = nxl_y, nxr_y |
|---|
| 778 | !$acc loop vector( 32 ) |
|---|
| 779 | DO j = 0, ny |
|---|
| 780 | f_out(j,i,k) = f_inv(i,k,j) |
|---|
| 781 | ENDDO |
|---|
| 782 | ENDDO |
|---|
| 783 | ENDDO |
|---|
| 784 | !$acc end kernels |
|---|
| 785 | !$OMP END PARALLEL |
|---|
| 786 | |
|---|
| 787 | END SUBROUTINE transpose_zy |
|---|
| 788 | |
|---|
| 789 | |
|---|
| 790 | SUBROUTINE transpose_zyd( f_in, work, f_out ) |
|---|
| 791 | |
|---|
| 792 | !------------------------------------------------------------------------------! |
|---|
| 793 | ! Description: |
|---|
| 794 | ! ------------ |
|---|
| 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 | ! This is a direct transposition for arrays with indices in regular order |
|---|
| 799 | ! (k,j,i) (cf. transpose_zy). |
|---|
| 800 | !------------------------------------------------------------------------------! |
|---|
| 801 | |
|---|
| 802 | USE cpulog |
|---|
| 803 | USE indices |
|---|
| 804 | USE interfaces |
|---|
| 805 | USE pegrid |
|---|
| 806 | USE transpose_indices |
|---|
| 807 | |
|---|
| 808 | IMPLICIT NONE |
|---|
| 809 | |
|---|
| 810 | INTEGER :: i, j, k, l, m, ys |
|---|
| 811 | |
|---|
| 812 | REAL :: f_in(1:nz,nys:nyn,nxl:nxr), f_inv(nys:nyn,nxl:nxr,1:nz), & |
|---|
| 813 | f_out(0:ny,nxl_yd:nxr_yd,nzb_yd:nzt_yd), & |
|---|
| 814 | work(nnx*nny*nnz) |
|---|
| 815 | |
|---|
| 816 | #if defined( __parallel ) |
|---|
| 817 | |
|---|
| 818 | ! |
|---|
| 819 | !-- Rearrange indices of input array in order to make data to be send |
|---|
| 820 | !-- by MPI contiguous |
|---|
| 821 | DO i = nxl, nxr |
|---|
| 822 | DO j = nys, nyn |
|---|
| 823 | DO k = 1, nz |
|---|
| 824 | f_inv(j,i,k) = f_in(k,j,i) |
|---|
| 825 | ENDDO |
|---|
| 826 | ENDDO |
|---|
| 827 | ENDDO |
|---|
| 828 | |
|---|
| 829 | ! |
|---|
| 830 | !-- Move data to different array, because memory location of work1 is |
|---|
| 831 | !-- needed further below (work1 = work2). |
|---|
| 832 | !-- If the PE grid is one-dimensional along x, only local reordering |
|---|
| 833 | !-- of the data is necessary and no transposition has to be done. |
|---|
| 834 | IF ( pdims(2) == 1 ) THEN |
|---|
| 835 | DO k = 1, nz |
|---|
| 836 | DO i = nxl, nxr |
|---|
| 837 | DO j = nys, nyn |
|---|
| 838 | f_out(j,i,k) = f_inv(j,i,k) |
|---|
| 839 | ENDDO |
|---|
| 840 | ENDDO |
|---|
| 841 | ENDDO |
|---|
| 842 | RETURN |
|---|
| 843 | ENDIF |
|---|
| 844 | |
|---|
| 845 | ! |
|---|
| 846 | !-- Transpose array |
|---|
| 847 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' ) |
|---|
| 848 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 849 | CALL MPI_ALLTOALL( f_inv(nys,nxl,1), sendrecvcount_zyd, MPI_REAL, & |
|---|
| 850 | work(1), sendrecvcount_zyd, MPI_REAL, & |
|---|
| 851 | comm1dy, ierr ) |
|---|
| 852 | CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'stop' ) |
|---|
| 853 | |
|---|
| 854 | ! |
|---|
| 855 | !-- Reorder transposed array |
|---|
| 856 | m = 0 |
|---|
| 857 | DO l = 0, pdims(2) - 1 |
|---|
| 858 | ys = 0 + l * nny |
|---|
| 859 | DO k = nzb_yd, nzt_yd |
|---|
| 860 | DO i = nxl_yd, nxr_yd |
|---|
| 861 | DO j = ys, ys + nny - 1 |
|---|
| 862 | m = m + 1 |
|---|
| 863 | f_out(j,i,k) = work(m) |
|---|
| 864 | ENDDO |
|---|
| 865 | ENDDO |
|---|
| 866 | ENDDO |
|---|
| 867 | ENDDO |
|---|
| 868 | |
|---|
| 869 | #endif |
|---|
| 870 | |
|---|
| 871 | END SUBROUTINE transpose_zyd |
|---|
