2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2009,2010,2012,2013,2014, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
44 #define GMX_PARALLEL_ENV_INITIALIZED 1
47 #define GMX_PARALLEL_ENV_INITIALIZED 1
49 #define GMX_PARALLEL_ENV_INITIALIZED 0
53 #include "gromacs/utility/gmxmpi.h"
56 /* TODO: Do we still need this? Are we still planning ot use fftw + OpenMP? */
58 /* requires fftw compiled with openmp */
59 /* #define FFT5D_FFTW_THREADS (now set by cmake) */
66 #include "gromacs/utility/smalloc.h"
68 #ifndef __FLT_EPSILON__
69 #define __FLT_EPSILON__ FLT_EPSILON
70 #define __DBL_EPSILON__ DBL_EPSILON
77 #include "gromacs/utility/fatalerror.h"
81 #include "thread_mpi/mutex.h"
82 #include "gromacs/utility/exceptions.h"
83 /* none of the fftw3 calls, except execute(), are thread-safe, so
84 we need to serialize them with this mutex. */
85 static tMPI::mutex big_fftw_mutex;
86 #define FFTW_LOCK try { big_fftw_mutex.lock(); } GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
87 #define FFTW_UNLOCK try { big_fftw_mutex.unlock(); } GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
88 #endif /* GMX_FFT_FFTW3 */
90 /* largest factor smaller than sqrt */
91 static int lfactor(int z)
94 for (i = static_cast<int>(sqrt(static_cast<double>(z)));; i--)
105 static int l2factor(int z)
122 /* largest prime factor: WARNING: slow recursion, only use for small numbers */
123 static int lpfactor(int z)
130 return std::max(lpfactor(f), lpfactor(z/f));
134 #ifdef HAVE_GETTIMEOFDAY
135 #include <sys/time.h>
139 gettimeofday(&tv, 0);
140 return tv.tv_sec+tv.tv_usec*1e-6;
150 static int vmax(int* a, int s)
153 for (i = 0; i < s; i++)
164 /* NxMxK the size of the data
165 * comm communicator to use for fft5d
166 * P0 number of processor in 1st axes (can be null for automatic)
167 * lin is allocated by fft5d because size of array is only known after planning phase
168 * rlout2 is only used as intermediate buffer - only returned after allocation to reuse for back transform - should not be used by caller
170 fft5d_plan fft5d_plan_3d(int NG, int MG, int KG, MPI_Comm comm[2], int flags, t_complex** rlin, t_complex** rlout, t_complex** rlout2, t_complex** rlout3, int nthreads)
173 int P[2], bMaster, prank[2], i, t;
175 int *N0 = 0, *N1 = 0, *M0 = 0, *M1 = 0, *K0 = 0, *K1 = 0, *oN0 = 0, *oN1 = 0, *oM0 = 0, *oM1 = 0, *oK0 = 0, *oK1 = 0;
176 int N[3], M[3], K[3], pN[3], pM[3], pK[3], oM[3], oK[3], *iNin[3] = {0}, *oNin[3] = {0}, *iNout[3] = {0}, *oNout[3] = {0};
177 int C[3], rC[3], nP[2];
179 t_complex *lin = 0, *lout = 0, *lout2 = 0, *lout3 = 0;
183 /* comm, prank and P are in the order of the decomposition (plan->cart is in the order of transposes) */
185 if (GMX_PARALLEL_ENV_INITIALIZED && comm[0] != MPI_COMM_NULL)
187 MPI_Comm_size(comm[0], &P[0]);
188 MPI_Comm_rank(comm[0], &prank[0]);
197 if (GMX_PARALLEL_ENV_INITIALIZED && comm[1] != MPI_COMM_NULL)
199 MPI_Comm_size(comm[1], &P[1]);
200 MPI_Comm_rank(comm[1], &prank[1]);
209 bMaster = (prank[0] == 0 && prank[1] == 0);
214 fprintf(debug, "FFT5D: Using %dx%d rank grid, rank %d,%d\n",
215 P[0], P[1], prank[0], prank[1]);
222 fprintf(debug, "FFT5D: N: %d, M: %d, K: %d, P: %dx%d, real2complex: %d, backward: %d, order yz: %d, debug %d\n",
223 NG, MG, KG, P[0], P[1], (flags&FFT5D_REALCOMPLEX) > 0, (flags&FFT5D_BACKWARD) > 0, (flags&FFT5D_ORDER_YZ) > 0, (flags&FFT5D_DEBUG) > 0);
225 /* The check below is not correct, one prime factor 11 or 13 is ok.
226 if (fft5d_fmax(fft5d_fmax(lpfactor(NG),lpfactor(MG)),lpfactor(KG))>7) {
227 printf("WARNING: FFT very slow with prime factors larger 7\n");
228 printf("Change FFT size or in case you cannot change it look at\n");
229 printf("http://www.fftw.org/fftw3_doc/Generating-your-own-code.html\n");
234 if (NG == 0 || MG == 0 || KG == 0)
238 printf("FFT5D: FATAL: Datasize cannot be zero in any dimension\n");
243 rNG = NG; rMG = MG; rKG = KG;
245 if (flags&FFT5D_REALCOMPLEX)
247 if (!(flags&FFT5D_BACKWARD))
253 if (!(flags&FFT5D_ORDER_YZ))
265 /*for transpose we need to know the size for each processor not only our own size*/
267 N0 = (int*)malloc(P[0]*sizeof(int)); N1 = (int*)malloc(P[1]*sizeof(int));
268 M0 = (int*)malloc(P[0]*sizeof(int)); M1 = (int*)malloc(P[1]*sizeof(int));
269 K0 = (int*)malloc(P[0]*sizeof(int)); K1 = (int*)malloc(P[1]*sizeof(int));
270 oN0 = (int*)malloc(P[0]*sizeof(int)); oN1 = (int*)malloc(P[1]*sizeof(int));
271 oM0 = (int*)malloc(P[0]*sizeof(int)); oM1 = (int*)malloc(P[1]*sizeof(int));
272 oK0 = (int*)malloc(P[0]*sizeof(int)); oK1 = (int*)malloc(P[1]*sizeof(int));
274 for (i = 0; i < P[0]; i++)
278 oN0[i] = i*ceil((double)NG/P[0]);
279 oM0[i] = i*ceil((double)MG/P[0]);
280 oK0[i] = i*ceil((double)KG/P[0]);
282 oN0[i] = (NG*i)/P[0];
283 oM0[i] = (MG*i)/P[0];
284 oK0[i] = (KG*i)/P[0];
287 for (i = 0; i < P[1]; i++)
290 oN1[i] = i*ceil((double)NG/P[1]);
291 oM1[i] = i*ceil((double)MG/P[1]);
292 oK1[i] = i*ceil((double)KG/P[1]);
294 oN1[i] = (NG*i)/P[1];
295 oM1[i] = (MG*i)/P[1];
296 oK1[i] = (KG*i)/P[1];
299 for (i = 0; i < P[0]-1; i++)
301 N0[i] = oN0[i+1]-oN0[i];
302 M0[i] = oM0[i+1]-oM0[i];
303 K0[i] = oK0[i+1]-oK0[i];
305 N0[P[0]-1] = NG-oN0[P[0]-1];
306 M0[P[0]-1] = MG-oM0[P[0]-1];
307 K0[P[0]-1] = KG-oK0[P[0]-1];
308 for (i = 0; i < P[1]-1; i++)
310 N1[i] = oN1[i+1]-oN1[i];
311 M1[i] = oM1[i+1]-oM1[i];
312 K1[i] = oK1[i+1]-oK1[i];
314 N1[P[1]-1] = NG-oN1[P[1]-1];
315 M1[P[1]-1] = MG-oM1[P[1]-1];
316 K1[P[1]-1] = KG-oK1[P[1]-1];
318 /* for step 1-3 the local N,M,K sizes of the transposed system
319 C: contiguous dimension, and nP: number of processor in subcommunicator
323 pM[0] = M0[prank[0]];
324 oM[0] = oM0[prank[0]];
325 pK[0] = K1[prank[1]];
326 oK[0] = oK1[prank[1]];
329 if (!(flags&FFT5D_ORDER_YZ))
331 N[0] = vmax(N1, P[1]);
333 K[0] = vmax(K1, P[1]);
334 pN[0] = N1[prank[1]];
340 N[1] = vmax(K0, P[0]);
341 pN[1] = K0[prank[0]];
346 M[1] = vmax(M0, P[0]);
347 pM[1] = M0[prank[0]];
348 oM[1] = oM0[prank[0]];
350 pK[1] = N1[prank[1]];
351 oK[1] = oN1[prank[1]];
357 M[2] = vmax(K0, P[0]);
358 pM[2] = K0[prank[0]];
359 oM[2] = oK0[prank[0]];
360 K[2] = vmax(N1, P[1]);
361 pK[2] = N1[prank[1]];
362 oK[2] = oN1[prank[1]];
363 free(N0); free(oN0); /*these are not used for this order*/
364 free(M1); free(oM1); /*the rest is freed in destroy*/
368 N[0] = vmax(N0, P[0]);
369 M[0] = vmax(M0, P[0]);
371 pN[0] = N0[prank[0]];
377 N[1] = vmax(M1, P[1]);
378 pN[1] = M1[prank[1]];
384 pM[1] = N0[prank[0]];
385 oM[1] = oN0[prank[0]];
386 K[1] = vmax(K1, P[1]);
387 pK[1] = K1[prank[1]];
388 oK[1] = oK1[prank[1]];
394 M[2] = vmax(N0, P[0]);
395 pM[2] = N0[prank[0]];
396 oM[2] = oN0[prank[0]];
397 K[2] = vmax(M1, P[1]);
398 pK[2] = M1[prank[1]];
399 oK[2] = oM1[prank[1]];
400 free(N1); free(oN1); /*these are not used for this order*/
401 free(K0); free(oK0); /*the rest is freed in destroy*/
403 N[2] = pN[2] = -1; /*not used*/
406 Difference between x-y-z regarding 2d decomposition is whether they are
407 distributed along axis 1, 2 or both
410 /* int lsize = fmax(N[0]*M[0]*K[0]*nP[0],N[1]*M[1]*K[1]*nP[1]); */
411 lsize = std::max(N[0]*M[0]*K[0]*nP[0], std::max(N[1]*M[1]*K[1]*nP[1], C[2]*M[2]*K[2]));
412 /* int lsize = fmax(C[0]*M[0]*K[0],fmax(C[1]*M[1]*K[1],C[2]*M[2]*K[2])); */
413 if (!(flags&FFT5D_NOMALLOC))
415 snew_aligned(lin, lsize, 32);
416 snew_aligned(lout, lsize, 32);
419 /* We need extra transpose buffers to avoid OpenMP barriers */
420 snew_aligned(lout2, lsize, 32);
421 snew_aligned(lout3, lsize, 32);
425 /* We can reuse the buffers to avoid cache misses */
446 plan = (fft5d_plan)calloc(1, sizeof(struct fft5d_plan_t));
451 fprintf(debug, "Running on %d threads\n", nthreads);
454 #ifdef GMX_FFT_FFTW3 /*if not FFTW - then we don't do a 3d plan but instead use only 1D plans */
455 /* It is possible to use the 3d plan with OMP threads - but in that case it is not allowed to be called from
456 * within a parallel region. For now deactivated. If it should be supported it has to made sure that
457 * that the execute of the 3d plan is in a master/serial block (since it contains it own parallel region)
458 * and that the 3d plan is faster than the 1d plan.
460 if ((!(flags&FFT5D_INPLACE)) && (!(P[0] > 1 || P[1] > 1)) && nthreads == 1) /*don't do 3d plan in parallel or if in_place requested */
462 int fftwflags = FFTW_DESTROY_INPUT;
464 int inNG = NG, outMG = MG, outKG = KG;
467 if (!(flags&FFT5D_NOMEASURE))
469 fftwflags |= FFTW_MEASURE;
471 if (flags&FFT5D_REALCOMPLEX)
473 if (!(flags&FFT5D_BACKWARD)) /*input pointer is not complex*/
477 else /*output pointer is not complex*/
479 if (!(flags&FFT5D_ORDER_YZ))
490 if (!(flags&FFT5D_BACKWARD))
496 dims[0].is = inNG*MG; /*N M K*/
499 if (!(flags&FFT5D_ORDER_YZ))
501 dims[0].os = MG; /*M K N*/
507 dims[0].os = 1; /*K N M*/
514 if (!(flags&FFT5D_ORDER_YZ))
524 dims[0].os = outMG*KG;
538 dims[0].os = outKG*NG;
544 #ifdef FFT5D_FFTW_THREADS
545 FFTW(plan_with_nthreads)(nthreads);
548 if ((flags&FFT5D_REALCOMPLEX) && !(flags&FFT5D_BACKWARD))
550 plan->p3d = FFTW(plan_guru_dft_r2c)(/*rank*/ 3, dims,
551 /*howmany*/ 0, /*howmany_dims*/ 0,
552 (real*)lin, (FFTW(complex) *) lout,
553 /*flags*/ fftwflags);
555 else if ((flags&FFT5D_REALCOMPLEX) && (flags&FFT5D_BACKWARD))
557 plan->p3d = FFTW(plan_guru_dft_c2r)(/*rank*/ 3, dims,
558 /*howmany*/ 0, /*howmany_dims*/ 0,
559 (FFTW(complex) *) lin, (real*)lout,
560 /*flags*/ fftwflags);
564 plan->p3d = FFTW(plan_guru_dft)(/*rank*/ 3, dims,
565 /*howmany*/ 0, /*howmany_dims*/ 0,
566 (FFTW(complex) *) lin, (FFTW(complex) *) lout,
567 /*sign*/ (flags&FFT5D_BACKWARD) ? 1 : -1, /*flags*/ fftwflags);
570 #ifdef FFT5D_FFTW_THREADS
571 FFTW(plan_with_nthreads)(1);
576 if (!plan->p3d) /* for decomposition and if 3d plan did not work */
578 #endif /* GMX_FFT_FFTW3 */
579 for (s = 0; s < 3; s++)
583 fprintf(debug, "FFT5D: Plan s %d rC %d M %d pK %d C %d lsize %d\n",
584 s, rC[s], M[s], pK[s], C[s], lsize);
586 plan->p1d[s] = (gmx_fft_t*)malloc(sizeof(gmx_fft_t)*nthreads);
588 /* Make sure that the init routines are only called by one thread at a time and in order
589 (later is only important to not confuse valgrind)
591 #pragma omp parallel for num_threads(nthreads) schedule(static) ordered
592 for (t = 0; t < nthreads; t++)
596 int tsize = ((t+1)*pM[s]*pK[s]/nthreads)-(t*pM[s]*pK[s]/nthreads);
598 if ((flags&FFT5D_REALCOMPLEX) && ((!(flags&FFT5D_BACKWARD) && s == 0) || ((flags&FFT5D_BACKWARD) && s == 2)))
600 gmx_fft_init_many_1d_real( &plan->p1d[s][t], rC[s], tsize, (flags&FFT5D_NOMEASURE) ? GMX_FFT_FLAG_CONSERVATIVE : 0 );
604 gmx_fft_init_many_1d ( &plan->p1d[s][t], C[s], tsize, (flags&FFT5D_NOMEASURE) ? GMX_FFT_FLAG_CONSERVATIVE : 0 );
613 if ((flags&FFT5D_ORDER_YZ)) /*plan->cart is in the order of transposes */
615 plan->cart[0] = comm[0]; plan->cart[1] = comm[1];
619 plan->cart[1] = comm[0]; plan->cart[0] = comm[1];
621 #ifdef FFT5D_MPI_TRANSPOSE
623 for (s = 0; s < 2; s++)
625 if ((s == 0 && !(flags&FFT5D_ORDER_YZ)) || (s == 1 && (flags&FFT5D_ORDER_YZ)))
627 plan->mpip[s] = FFTW(mpi_plan_many_transpose)(nP[s], nP[s], N[s]*K[s]*pM[s]*2, 1, 1, (real*)lout2, (real*)lout3, plan->cart[s], FFTW_PATIENT);
631 plan->mpip[s] = FFTW(mpi_plan_many_transpose)(nP[s], nP[s], N[s]*pK[s]*M[s]*2, 1, 1, (real*)lout2, (real*)lout3, plan->cart[s], FFTW_PATIENT);
643 plan->NG = NG; plan->MG = MG; plan->KG = KG;
645 for (s = 0; s < 3; s++)
647 plan->N[s] = N[s]; plan->M[s] = M[s]; plan->K[s] = K[s]; plan->pN[s] = pN[s]; plan->pM[s] = pM[s]; plan->pK[s] = pK[s];
648 plan->oM[s] = oM[s]; plan->oK[s] = oK[s];
649 plan->C[s] = C[s]; plan->rC[s] = rC[s];
650 plan->iNin[s] = iNin[s]; plan->oNin[s] = oNin[s]; plan->iNout[s] = iNout[s]; plan->oNout[s] = oNout[s];
652 for (s = 0; s < 2; s++)
654 plan->P[s] = nP[s]; plan->coor[s] = prank[s];
657 /* plan->fftorder=fftorder;
658 plan->direction=direction;
659 plan->realcomplex=realcomplex;
662 plan->nthreads = nthreads;
682 /*here x,y,z and N,M,K is in rotated coordinate system!!
683 x (and N) is mayor (consecutive) dimension, y (M) middle and z (K) major
684 maxN,maxM,maxK is max size of local data
685 pN, pM, pK is local size specific to current processor (only different to max if not divisible)
686 NG, MG, KG is size of global data*/
687 static void splitaxes(t_complex* lout, const t_complex* lin,
688 int maxN, int maxM, int maxK, int pM,
689 int P, int NG, int *N, int* oN, int starty, int startz, int endy, int endz)
692 int in_i, out_i, in_z, out_z, in_y, out_y;
695 for (z = startz; z < endz+1; z++) /*3. z l*/
716 for (i = 0; i < P; i++) /*index cube along long axis*/
718 out_i = out_z + i*maxN*maxM*maxK;
720 for (y = s_y; y < e_y; y++) /*2. y k*/
722 out_y = out_i + y*maxN;
724 for (x = 0; x < N[i]; x++) /*1. x j*/
726 lout[out_y+x] = lin[in_y+x]; /*in=z*NG*pM+oN[i]+y*NG+x*/
727 /*after split important that each processor chunk i has size maxN*maxM*maxK and thus being the same size*/
728 /*before split data contiguos - thus if different processor get different amount oN is different*/
735 /*make axis contiguous again (after AllToAll) and also do local transpose*/
736 /*transpose mayor and major dimension
738 the major, middle, minor order is only correct for x,y,z (N,M,K) for the input
739 N,M,K local dimensions
741 static void joinAxesTrans13(t_complex* lout, const t_complex* lin,
742 int maxN, int maxM, int maxK, int pM,
743 int P, int KG, int* K, int* oK, int starty, int startx, int endy, int endx)
746 int out_i, in_i, out_x, in_x, out_z, in_z;
749 for (x = startx; x < endx+1; x++) /*1.j*/
771 for (i = 0; i < P; i++) /*index cube along long axis*/
773 out_i = out_x + oK[i];
774 in_i = in_x + i*maxM*maxN*maxK;
775 for (z = 0; z < K[i]; z++) /*3.l*/
778 in_z = in_i + z*maxM*maxN;
779 for (y = s_y; y < e_y; y++) /*2.k*/
781 lout[out_z+y*KG] = lin[in_z+y*maxN]; /*out=x*KG*pM+oK[i]+z+y*KG*/
788 /*make axis contiguous again (after AllToAll) and also do local transpose
789 tranpose mayor and middle dimension
791 the minor, middle, major order is only correct for x,y,z (N,M,K) for the input
794 static void joinAxesTrans12(t_complex* lout, const t_complex* lin, int maxN, int maxM, int maxK, int pN,
795 int P, int MG, int* M, int* oM, int startx, int startz, int endx, int endz)
798 int out_i, in_i, out_z, in_z, out_x, in_x;
801 for (z = startz; z < endz+1; z++)
822 for (i = 0; i < P; i++) /*index cube along long axis*/
824 out_i = out_z + oM[i];
825 in_i = in_z + i*maxM*maxN*maxK;
826 for (x = s_x; x < e_x; x++)
828 out_x = out_i + x*MG;
830 for (y = 0; y < M[i]; y++)
832 lout[out_x+y] = lin[in_x+y*maxN]; /*out=z*MG*pN+oM[i]+x*MG+y*/
840 static void rotate_offsets(int x[])
851 /*compute the offset to compare or print transposed local data in original input coordinates
852 xs matrix dimension size, xl dimension length, xc decomposition offset
853 s: step in computation = number of transposes*/
854 static void compute_offsets(fft5d_plan plan, int xs[], int xl[], int xc[], int NG[], int s)
856 /* int direction = plan->direction;
857 int fftorder = plan->fftorder;*/
861 int *pM = plan->pM, *pK = plan->pK, *oM = plan->oM, *oK = plan->oK,
862 *C = plan->C, *rC = plan->rC;
864 NG[0] = plan->NG; NG[1] = plan->MG; NG[2] = plan->KG;
866 if (!(plan->flags&FFT5D_ORDER_YZ))
870 case 0: o = XYZ; break;
871 case 1: o = ZYX; break;
872 case 2: o = YZX; break;
880 case 0: o = XYZ; break;
881 case 1: o = YXZ; break;
882 case 2: o = ZXY; break;
889 case XYZ: pos[0] = 1; pos[1] = 2; pos[2] = 3; break;
890 case XZY: pos[0] = 1; pos[1] = 3; pos[2] = 2; break;
891 case YXZ: pos[0] = 2; pos[1] = 1; pos[2] = 3; break;
892 case YZX: pos[0] = 3; pos[1] = 1; pos[2] = 2; break;
893 case ZXY: pos[0] = 2; pos[1] = 3; pos[2] = 1; break;
894 case ZYX: pos[0] = 3; pos[1] = 2; pos[2] = 1; break;
896 /*if (debug) printf("pos: %d %d %d\n",pos[0],pos[1],pos[2]);*/
898 /*xs, xl give dimension size and data length in local transposed coordinate system
899 for 0(/1/2): x(/y/z) in original coordinate system*/
900 for (i = 0; i < 3; i++)
904 case 1: xs[i] = 1; xc[i] = 0; xl[i] = C[s]; break;
905 case 2: xs[i] = C[s]; xc[i] = oM[s]; xl[i] = pM[s]; break;
906 case 3: xs[i] = C[s]*pM[s]; xc[i] = oK[s]; xl[i] = pK[s]; break;
909 /*input order is different for test program to match FFTW order
910 (important for complex to real)*/
911 if (plan->flags&FFT5D_BACKWARD)
917 if (plan->flags&FFT5D_ORDER_YZ)
925 if ((plan->flags&FFT5D_REALCOMPLEX) && ((!(plan->flags&FFT5D_BACKWARD) && s == 0) || ((plan->flags&FFT5D_BACKWARD) && s == 2)))
931 static void print_localdata(const t_complex* lin, const char* txt, int s, fft5d_plan plan)
934 int *coor = plan->coor;
935 int xs[3], xl[3], xc[3], NG[3];
936 int ll = (plan->flags&FFT5D_REALCOMPLEX) ? 1 : 2;
937 compute_offsets(plan, xs, xl, xc, NG, s);
938 fprintf(debug, txt, coor[0], coor[1], s);
939 /*printf("xs: %d %d %d, xl: %d %d %d\n",xs[0],xs[1],xs[2],xl[0],xl[1],xl[2]);*/
940 for (z = 0; z < xl[2]; z++)
942 for (y = 0; y < xl[1]; y++)
944 fprintf(debug, "%d %d: ", coor[0], coor[1]);
945 for (x = 0; x < xl[0]; x++)
947 for (l = 0; l < ll; l++)
949 fprintf(debug, "%f ", ((real*)lin)[(z*xs[2]+y*xs[1])*2+(x*xs[0])*ll+l]);
953 fprintf(debug, "\n");
958 void fft5d_execute(fft5d_plan plan, int thread, fft5d_time times)
960 t_complex *lin = plan->lin;
961 t_complex *lout = plan->lout;
962 t_complex *lout2 = plan->lout2;
963 t_complex *lout3 = plan->lout3;
964 t_complex *fftout, *joinin;
966 gmx_fft_t **p1d = plan->p1d;
967 #ifdef FFT5D_MPI_TRANSPOSE
968 FFTW(plan) *mpip = plan->mpip;
971 MPI_Comm *cart = plan->cart;
974 double time_fft = 0, time_local = 0, time_mpi[2] = {0}, time = 0;
976 int *N = plan->N, *M = plan->M, *K = plan->K, *pN = plan->pN, *pM = plan->pM, *pK = plan->pK,
977 *C = plan->C, *P = plan->P, **iNin = plan->iNin, **oNin = plan->oNin, **iNout = plan->iNout, **oNout = plan->oNout;
978 int s = 0, tstart, tend, bParallelDim;
992 FFTW(execute)(plan->p3d);
996 times->fft += MPI_Wtime()-time;
1007 if (plan->flags&FFT5D_DEBUG && thread == 0)
1009 print_localdata(lin, "%d %d: copy in lin\n", s, plan);
1012 for (s = 0; s < 2; s++) /*loop over first two FFT steps (corner rotations)*/
1016 if (GMX_PARALLEL_ENV_INITIALIZED && cart[s] != MPI_COMM_NULL && P[s] > 1)
1026 /* ---------- START FFT ------------ */
1028 if (times != 0 && thread == 0)
1034 if (bParallelDim || plan->nthreads == 1)
1050 tstart = (thread*pM[s]*pK[s]/plan->nthreads)*C[s];
1051 if ((plan->flags&FFT5D_REALCOMPLEX) && !(plan->flags&FFT5D_BACKWARD) && s == 0)
1053 gmx_fft_many_1d_real(p1d[s][thread], (plan->flags&FFT5D_BACKWARD) ? GMX_FFT_COMPLEX_TO_REAL : GMX_FFT_REAL_TO_COMPLEX, lin+tstart, fftout+tstart);
1057 gmx_fft_many_1d( p1d[s][thread], (plan->flags&FFT5D_BACKWARD) ? GMX_FFT_BACKWARD : GMX_FFT_FORWARD, lin+tstart, fftout+tstart);
1062 if (times != NULL && thread == 0)
1064 time_fft += MPI_Wtime()-time;
1067 if (plan->flags&FFT5D_DEBUG && thread == 0)
1069 print_localdata(lout, "%d %d: FFT %d\n", s, plan);
1071 /* ---------- END FFT ------------ */
1073 /* ---------- START SPLIT + TRANSPOSE------------ (if parallel in in this dimension)*/
1077 if (times != NULL && thread == 0)
1084 1. (most outer) axes (x) is split into P[s] parts of size N[s]
1088 tend = ((thread+1)*pM[s]*pK[s]/plan->nthreads);
1090 splitaxes(lout2, lout, N[s], M[s], K[s], pM[s], P[s], C[s], iNout[s], oNout[s], tstart%pM[s], tstart/pM[s], tend%pM[s], tend/pM[s]);
1092 #pragma omp barrier /*barrier required before AllToAll (all input has to be their) - before timing to make timing more acurate*/
1094 if (times != NULL && thread == 0)
1096 time_local += MPI_Wtime()-time;
1100 /* ---------- END SPLIT , START TRANSPOSE------------ */
1110 wallcycle_start(times, ewcPME_FFTCOMM);
1112 #ifdef FFT5D_MPI_TRANSPOSE
1113 FFTW(execute)(mpip[s]);
1116 if ((s == 0 && !(plan->flags&FFT5D_ORDER_YZ)) || (s == 1 && (plan->flags&FFT5D_ORDER_YZ)))
1118 MPI_Alltoall((real *)lout2, N[s]*pM[s]*K[s]*sizeof(t_complex)/sizeof(real), GMX_MPI_REAL, (real *)lout3, N[s]*pM[s]*K[s]*sizeof(t_complex)/sizeof(real), GMX_MPI_REAL, cart[s]);
1122 MPI_Alltoall((real *)lout2, N[s]*M[s]*pK[s]*sizeof(t_complex)/sizeof(real), GMX_MPI_REAL, (real *)lout3, N[s]*M[s]*pK[s]*sizeof(t_complex)/sizeof(real), GMX_MPI_REAL, cart[s]);
1125 gmx_incons("fft5d MPI call without MPI configuration");
1127 #endif /*FFT5D_MPI_TRANSPOSE*/
1131 time_mpi[s] = MPI_Wtime()-time;
1134 wallcycle_stop(times, ewcPME_FFTCOMM);
1138 #pragma omp barrier /*both needed for parallel and non-parallel dimension (either have to wait on data from AlltoAll or from last FFT*/
1140 /* ---------- END SPLIT + TRANSPOSE------------ */
1142 /* ---------- START JOIN ------------ */
1144 if (times != NULL && thread == 0)
1158 /*bring back in matrix form
1159 thus make new 1. axes contiguos
1160 also local transpose 1 and 2/3
1161 runs on thread used for following FFT (thus needing a barrier before but not afterwards)
1163 if ((s == 0 && !(plan->flags&FFT5D_ORDER_YZ)) || (s == 1 && (plan->flags&FFT5D_ORDER_YZ)))
1167 tstart = ( thread *pM[s]*pN[s]/plan->nthreads);
1168 tend = ((thread+1)*pM[s]*pN[s]/plan->nthreads);
1169 joinAxesTrans13(lin, joinin, N[s], pM[s], K[s], pM[s], P[s], C[s+1], iNin[s+1], oNin[s+1], tstart%pM[s], tstart/pM[s], tend%pM[s], tend/pM[s]);
1176 tstart = ( thread *pK[s]*pN[s]/plan->nthreads);
1177 tend = ((thread+1)*pK[s]*pN[s]/plan->nthreads);
1178 joinAxesTrans12(lin, joinin, N[s], M[s], pK[s], pN[s], P[s], C[s+1], iNin[s+1], oNin[s+1], tstart%pN[s], tstart/pN[s], tend%pN[s], tend/pN[s]);
1183 if (times != NULL && thread == 0)
1185 time_local += MPI_Wtime()-time;
1188 if (plan->flags&FFT5D_DEBUG && thread == 0)
1190 print_localdata(lin, "%d %d: tranposed %d\n", s+1, plan);
1192 /* ---------- END JOIN ------------ */
1194 /*if (debug) print_localdata(lin, "%d %d: transposed x-z\n", N1, M0, K, ZYX, coor);*/
1195 } /* for(s=0;s<2;s++) */
1197 if (times != NULL && thread == 0)
1203 if (plan->flags&FFT5D_INPLACE)
1205 lout = lin; /*in place currently not supported*/
1208 /* ----------- FFT ----------- */
1209 tstart = (thread*pM[s]*pK[s]/plan->nthreads)*C[s];
1210 if ((plan->flags&FFT5D_REALCOMPLEX) && (plan->flags&FFT5D_BACKWARD))
1212 gmx_fft_many_1d_real(p1d[s][thread], (plan->flags&FFT5D_BACKWARD) ? GMX_FFT_COMPLEX_TO_REAL : GMX_FFT_REAL_TO_COMPLEX, lin+tstart, lout+tstart);
1216 gmx_fft_many_1d( p1d[s][thread], (plan->flags&FFT5D_BACKWARD) ? GMX_FFT_BACKWARD : GMX_FFT_FORWARD, lin+tstart, lout+tstart);
1218 /* ------------ END FFT ---------*/
1221 if (times != NULL && thread == 0)
1223 time_fft += MPI_Wtime()-time;
1225 times->fft += time_fft;
1226 times->local += time_local;
1227 times->mpi2 += time_mpi[1];
1228 times->mpi1 += time_mpi[0];
1232 if (plan->flags&FFT5D_DEBUG && thread == 0)
1234 print_localdata(lout, "%d %d: FFT %d\n", s, plan);
1238 void fft5d_destroy(fft5d_plan plan)
1242 for (s = 0; s < 3; s++)
1246 for (t = 0; t < plan->nthreads; t++)
1248 gmx_many_fft_destroy(plan->p1d[s][t]);
1254 free(plan->iNin[s]);
1259 free(plan->oNin[s]);
1264 free(plan->iNout[s]);
1269 free(plan->oNout[s]);
1273 #ifdef GMX_FFT_FFTW3
1275 #ifdef FFT5D_MPI_TRANSPOS
1276 for (s = 0; s < 2; s++)
1278 FFTW(destroy_plan)(plan->mpip[s]);
1280 #endif /* FFT5D_MPI_TRANSPOS */
1283 FFTW(destroy_plan)(plan->p3d);
1286 #endif /* GMX_FFT_FFTW3 */
1288 if (!(plan->flags&FFT5D_NOMALLOC))
1290 sfree_aligned(plan->lin);
1291 sfree_aligned(plan->lout);
1292 if (plan->nthreads > 1)
1294 sfree_aligned(plan->lout2);
1295 sfree_aligned(plan->lout3);
1299 #ifdef FFT5D_THREADS
1300 #ifdef FFT5D_FFTW_THREADS
1301 /*FFTW(cleanup_threads)();*/
1308 /*Is this better than direct access of plan? enough data?
1309 here 0,1 reference divided by which processor grid dimension (not FFT step!)*/
1310 void fft5d_local_size(fft5d_plan plan, int* N1, int* M0, int* K0, int* K1, int** coor)
1321 /*same as fft5d_plan_3d but with cartesian coordinator and automatic splitting
1322 of processor dimensions*/
1323 fft5d_plan fft5d_plan_3d_cart(int NG, int MG, int KG, MPI_Comm comm, int P0, int flags, t_complex** rlin, t_complex** rlout, t_complex** rlout2, t_complex** rlout3, int nthreads)
1325 MPI_Comm cart[2] = {0};
1327 int size = 1, prank = 0;
1330 int wrap[] = {0, 0};
1332 int rdim1[] = {0, 1}, rdim2[] = {1, 0};
1334 MPI_Comm_size(comm, &size);
1335 MPI_Comm_rank(comm, &prank);
1345 printf("FFT5D: WARNING: Number of ranks %d not evenly divisible by %d\n", size, P0);
1350 P[0] = P0; P[1] = size/P0; /*number of processors in the two dimensions*/
1352 /*Difference between x-y-z regarding 2d decomposition is whether they are
1353 distributed along axis 1, 2 or both*/
1355 MPI_Cart_create(comm, 2, P, wrap, 1, &gcart); /*parameter 4: value 1: reorder*/
1356 MPI_Cart_get(gcart, 2, P, wrap, coor);
1357 MPI_Cart_sub(gcart, rdim1, &cart[0]);
1358 MPI_Cart_sub(gcart, rdim2, &cart[1]);
1363 return fft5d_plan_3d(NG, MG, KG, cart, flags, rlin, rlout, rlout2, rlout3, nthreads);
1368 /*prints in original coordinate system of data (as the input to FFT)*/
1369 void fft5d_compare_data(const t_complex* lin, const t_complex* in, fft5d_plan plan, int bothLocal, int normalize)
1371 int xs[3], xl[3], xc[3], NG[3];
1373 int *coor = plan->coor;
1374 int ll = 2; /*compare ll values per element (has to be 2 for complex)*/
1375 if ((plan->flags&FFT5D_REALCOMPLEX) && (plan->flags&FFT5D_BACKWARD))
1380 compute_offsets(plan, xs, xl, xc, NG, 2);
1381 if (plan->flags&FFT5D_DEBUG)
1383 printf("Compare2\n");
1385 for (z = 0; z < xl[2]; z++)
1387 for (y = 0; y < xl[1]; y++)
1389 if (plan->flags&FFT5D_DEBUG)
1391 printf("%d %d: ", coor[0], coor[1]);
1393 for (x = 0; x < xl[0]; x++)
1395 for (l = 0; l < ll; l++) /*loop over real/complex parts*/
1398 a = ((real*)lin)[(z*xs[2]+y*xs[1])*2+x*xs[0]*ll+l];
1401 a /= plan->rC[0]*plan->rC[1]*plan->rC[2];
1405 b = ((real*)in)[((z+xc[2])*NG[0]*NG[1]+(y+xc[1])*NG[0])*2+(x+xc[0])*ll+l];
1409 b = ((real*)in)[(z*xs[2]+y*xs[1])*2+x*xs[0]*ll+l];
1411 if (plan->flags&FFT5D_DEBUG)
1413 printf("%f %f, ", a, b);
1417 if (fabs(a-b) > 2*NG[0]*NG[1]*NG[2]*GMX_REAL_EPS)
1419 printf("result incorrect on %d,%d at %d,%d,%d: FFT5D:%f reference:%f\n", coor[0], coor[1], x, y, z, a, b);
1421 /* assert(fabs(a-b)<2*NG[0]*NG[1]*NG[2]*GMX_REAL_EPS);*/
1424 if (plan->flags&FFT5D_DEBUG)
1429 if (plan->flags&FFT5D_DEBUG)