2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2005,2006,2007,2008,2009,2010,2011,2012,2013,2014, 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.
48 #include "gromacs/utility/smalloc.h"
49 #include "gmx_fatal.h"
50 #include "gmx_fatal_collective.h"
53 #include "domdec_network.h"
56 #include "chargegroup.h"
65 #include "mtop_util.h"
66 #include "gmx_ga2la.h"
68 #include "nbnxn_search.h"
70 #include "gmx_omp_nthreads.h"
71 #include "gpu_utils.h"
73 #include "gromacs/fileio/futil.h"
74 #include "gromacs/fileio/gmxfio.h"
75 #include "gromacs/fileio/pdbio.h"
76 #include "gromacs/timing/wallcycle.h"
77 #include "gromacs/utility/gmxmpi.h"
78 #include "gromacs/swap/swapcoords.h"
79 #include "gromacs/utility/qsort_threadsafe.h"
80 #include "gromacs/pulling/pull.h"
81 #include "gromacs/pulling/pull_rotation.h"
82 #include "gromacs/imd/imd.h"
84 #define DDRANK(dd, rank) (rank)
85 #define DDMASTERRANK(dd) (dd->masterrank)
87 typedef struct gmx_domdec_master
89 /* The cell boundaries */
91 /* The global charge group division */
92 int *ncg; /* Number of home charge groups for each node */
93 int *index; /* Index of nnodes+1 into cg */
94 int *cg; /* Global charge group index */
95 int *nat; /* Number of home atoms for each node. */
96 int *ibuf; /* Buffer for communication */
97 rvec *vbuf; /* Buffer for state scattering and gathering */
98 } gmx_domdec_master_t;
102 /* The numbers of charge groups to send and receive for each cell
103 * that requires communication, the last entry contains the total
104 * number of atoms that needs to be communicated.
106 int nsend[DD_MAXIZONE+2];
107 int nrecv[DD_MAXIZONE+2];
108 /* The charge groups to send */
111 /* The atom range for non-in-place communication */
112 int cell2at0[DD_MAXIZONE];
113 int cell2at1[DD_MAXIZONE];
118 int np; /* Number of grid pulses in this dimension */
119 int np_dlb; /* For dlb, for use with edlbAUTO */
120 gmx_domdec_ind_t *ind; /* The indices to communicate, size np */
122 gmx_bool bInPlace; /* Can we communicate in place? */
123 } gmx_domdec_comm_dim_t;
127 gmx_bool *bCellMin; /* Temp. var.: is this cell size at the limit */
128 real *cell_f; /* State var.: cell boundaries, box relative */
129 real *old_cell_f; /* Temp. var.: old cell size */
130 real *cell_f_max0; /* State var.: max lower boundary, incl neighbors */
131 real *cell_f_min1; /* State var.: min upper boundary, incl neighbors */
132 real *bound_min; /* Temp. var.: lower limit for cell boundary */
133 real *bound_max; /* Temp. var.: upper limit for cell boundary */
134 gmx_bool bLimited; /* State var.: is DLB limited in this dim and row */
135 real *buf_ncd; /* Temp. var. */
138 #define DD_NLOAD_MAX 9
140 /* Here floats are accurate enough, since these variables
141 * only influence the load balancing, not the actual MD results.
168 gmx_cgsort_t *sort_new;
180 /* This enum determines the order of the coordinates.
181 * ddnatHOME and ddnatZONE should be first and second,
182 * the others can be ordered as wanted.
185 ddnatHOME, ddnatZONE, ddnatVSITE, ddnatCON, ddnatNR
189 edlbAUTO, edlbNO, edlbYES, edlbNR
191 const char *edlb_names[edlbNR] = { "auto", "no", "yes" };
195 int dim; /* The dimension */
196 gmx_bool dim_match; /* Tells if DD and PME dims match */
197 int nslab; /* The number of PME slabs in this dimension */
198 real *slb_dim_f; /* Cell sizes for determining the PME comm. with SLB */
199 int *pp_min; /* The minimum pp node location, size nslab */
200 int *pp_max; /* The maximum pp node location,size nslab */
201 int maxshift; /* The maximum shift for coordinate redistribution in PME */
206 real min0; /* The minimum bottom of this zone */
207 real max1; /* The maximum top of this zone */
208 real min1; /* The minimum top of this zone */
209 real mch0; /* The maximum bottom communicaton height for this zone */
210 real mch1; /* The maximum top communicaton height for this zone */
211 real p1_0; /* The bottom value of the first cell in this zone */
212 real p1_1; /* The top value of the first cell in this zone */
217 gmx_domdec_ind_t ind;
224 } dd_comm_setup_work_t;
226 typedef struct gmx_domdec_comm
228 /* All arrays are indexed with 0 to dd->ndim (not Cartesian indexing),
229 * unless stated otherwise.
232 /* The number of decomposition dimensions for PME, 0: no PME */
234 /* The number of nodes doing PME (PP/PME or only PME) */
238 /* The communication setup including the PME only nodes */
239 gmx_bool bCartesianPP_PME;
242 int *pmenodes; /* size npmenodes */
243 int *ddindex2simnodeid; /* size npmenodes, only with bCartesianPP
244 * but with bCartesianPP_PME */
245 gmx_ddpme_t ddpme[2];
247 /* The DD particle-particle nodes only */
248 gmx_bool bCartesianPP;
249 int *ddindex2ddnodeid; /* size npmenode, only with bCartesianPP_PME */
251 /* The global charge groups */
254 /* Should we sort the cgs */
256 gmx_domdec_sort_t *sort;
258 /* Are there charge groups? */
261 /* Are there bonded and multi-body interactions between charge groups? */
262 gmx_bool bInterCGBondeds;
263 gmx_bool bInterCGMultiBody;
265 /* Data for the optional bonded interaction atom communication range */
272 /* Are we actually using DLB? */
273 gmx_bool bDynLoadBal;
275 /* Cell sizes for static load balancing, first index cartesian */
278 /* The width of the communicated boundaries */
281 /* The minimum cell size (including triclinic correction) */
283 /* For dlb, for use with edlbAUTO */
284 rvec cellsize_min_dlb;
285 /* The lower limit for the DD cell size with DLB */
287 /* Effectively no NB cut-off limit with DLB for systems without PBC? */
288 gmx_bool bVacDLBNoLimit;
290 /* With PME load balancing we set limits on DLB */
291 gmx_bool bPMELoadBalDLBLimits;
292 /* DLB needs to take into account that we want to allow this maximum
293 * cut-off (for PME load balancing), this could limit cell boundaries.
295 real PMELoadBal_max_cutoff;
297 /* tric_dir is only stored here because dd_get_ns_ranges needs it */
299 /* box0 and box_size are required with dim's without pbc and -gcom */
303 /* The cell boundaries */
307 /* The old location of the cell boundaries, to check cg displacements */
311 /* The communication setup and charge group boundaries for the zones */
312 gmx_domdec_zones_t zones;
314 /* The zone limits for DD dimensions 1 and 2 (not 0), determined from
315 * cell boundaries of neighboring cells for dynamic load balancing.
317 gmx_ddzone_t zone_d1[2];
318 gmx_ddzone_t zone_d2[2][2];
320 /* The coordinate/force communication setup and indices */
321 gmx_domdec_comm_dim_t cd[DIM];
322 /* The maximum number of cells to communicate with in one dimension */
325 /* Which cg distribution is stored on the master node */
326 int master_cg_ddp_count;
328 /* The number of cg's received from the direct neighbors */
329 int zone_ncg1[DD_MAXZONE];
331 /* The atom counts, the range for each type t is nat[t-1] <= at < nat[t] */
334 /* Array for signalling if atoms have moved to another domain */
338 /* Communication buffer for general use */
342 /* Communication buffer for general use */
345 /* Temporary storage for thread parallel communication setup */
347 dd_comm_setup_work_t *dth;
349 /* Communication buffers only used with multiple grid pulses */
354 /* Communication buffers for local redistribution */
356 int cggl_flag_nalloc[DIM*2];
358 int cgcm_state_nalloc[DIM*2];
360 /* Cell sizes for dynamic load balancing */
361 gmx_domdec_root_t **root;
365 real cell_f_max0[DIM];
366 real cell_f_min1[DIM];
368 /* Stuff for load communication */
369 gmx_bool bRecordLoad;
370 gmx_domdec_load_t *load;
371 int nrank_gpu_shared;
373 MPI_Comm *mpi_comm_load;
374 MPI_Comm mpi_comm_gpu_shared;
377 /* Maximum DLB scaling per load balancing step in percent */
381 float cycl[ddCyclNr];
382 int cycl_n[ddCyclNr];
383 float cycl_max[ddCyclNr];
384 /* Flop counter (0=no,1=yes,2=with (eFlop-1)*5% noise */
388 /* Have often have did we have load measurements */
390 /* Have often have we collected the load measurements */
394 double sum_nat[ddnatNR-ddnatZONE];
404 /* The last partition step */
405 gmx_int64_t partition_step;
413 /* The size per charge group of the cggl_flag buffer in gmx_domdec_comm_t */
416 /* The flags for the cggl_flag buffer in gmx_domdec_comm_t */
417 #define DD_FLAG_NRCG 65535
418 #define DD_FLAG_FW(d) (1<<(16+(d)*2))
419 #define DD_FLAG_BW(d) (1<<(16+(d)*2+1))
421 /* Zone permutation required to obtain consecutive charge groups
422 * for neighbor searching.
424 static const int zone_perm[3][4] = { {0, 0, 0, 0}, {1, 0, 0, 0}, {3, 0, 1, 2} };
426 /* dd_zo and dd_zp3/dd_zp2 are set up such that i zones with non-zero
427 * components see only j zones with that component 0.
430 /* The DD zone order */
431 static const ivec dd_zo[DD_MAXZONE] =
432 {{0, 0, 0}, {1, 0, 0}, {1, 1, 0}, {0, 1, 0}, {0, 1, 1}, {0, 0, 1}, {1, 0, 1}, {1, 1, 1}};
437 static const ivec dd_zp3[dd_zp3n] = {{0, 0, 8}, {1, 3, 6}, {2, 5, 6}, {3, 5, 7}};
442 static const ivec dd_zp2[dd_zp2n] = {{0, 0, 4}, {1, 3, 4}};
447 static const ivec dd_zp1[dd_zp1n] = {{0, 0, 2}};
449 /* Factors used to avoid problems due to rounding issues */
450 #define DD_CELL_MARGIN 1.0001
451 #define DD_CELL_MARGIN2 1.00005
452 /* Factor to account for pressure scaling during nstlist steps */
453 #define DD_PRES_SCALE_MARGIN 1.02
455 /* Turn on DLB when the load imbalance causes this amount of total loss.
456 * There is a bit of overhead with DLB and it's difficult to achieve
457 * a load imbalance of less than 2% with DLB.
459 #define DD_PERF_LOSS_DLB_ON 0.02
461 /* Warn about imbalance due to PP or PP/PME load imbalance at this loss */
462 #define DD_PERF_LOSS_WARN 0.05
464 #define DD_CELL_F_SIZE(dd, di) ((dd)->nc[(dd)->dim[(di)]]+1+(di)*2+1+(di))
466 /* Use separate MPI send and receive commands
467 * when nnodes <= GMX_DD_NNODES_SENDRECV.
468 * This saves memory (and some copying for small nnodes).
469 * For high parallelization scatter and gather calls are used.
471 #define GMX_DD_NNODES_SENDRECV 4
475 #define dd_index(n,i) ((((i)[ZZ]*(n)[YY] + (i)[YY])*(n)[XX]) + (i)[XX])
477 static void index2xyz(ivec nc,int ind,ivec xyz)
479 xyz[XX] = ind % nc[XX];
480 xyz[YY] = (ind / nc[XX]) % nc[YY];
481 xyz[ZZ] = ind / (nc[YY]*nc[XX]);
485 /* This order is required to minimize the coordinate communication in PME
486 * which uses decomposition in the x direction.
488 #define dd_index(n, i) ((((i)[XX]*(n)[YY] + (i)[YY])*(n)[ZZ]) + (i)[ZZ])
490 static void ddindex2xyz(ivec nc, int ind, ivec xyz)
492 xyz[XX] = ind / (nc[YY]*nc[ZZ]);
493 xyz[YY] = (ind / nc[ZZ]) % nc[YY];
494 xyz[ZZ] = ind % nc[ZZ];
497 static int ddcoord2ddnodeid(gmx_domdec_t *dd, ivec c)
502 ddindex = dd_index(dd->nc, c);
503 if (dd->comm->bCartesianPP_PME)
505 ddnodeid = dd->comm->ddindex2ddnodeid[ddindex];
507 else if (dd->comm->bCartesianPP)
510 MPI_Cart_rank(dd->mpi_comm_all, c, &ddnodeid);
521 static gmx_bool dynamic_dd_box(gmx_ddbox_t *ddbox, t_inputrec *ir)
523 return (ddbox->nboundeddim < DIM || DYNAMIC_BOX(*ir));
526 int ddglatnr(gmx_domdec_t *dd, int i)
536 if (i >= dd->comm->nat[ddnatNR-1])
538 gmx_fatal(FARGS, "glatnr called with %d, which is larger than the local number of atoms (%d)", i, dd->comm->nat[ddnatNR-1]);
540 atnr = dd->gatindex[i] + 1;
546 t_block *dd_charge_groups_global(gmx_domdec_t *dd)
548 return &dd->comm->cgs_gl;
551 static void vec_rvec_init(vec_rvec_t *v)
557 static void vec_rvec_check_alloc(vec_rvec_t *v, int n)
561 v->nalloc = over_alloc_dd(n);
562 srenew(v->v, v->nalloc);
566 void dd_store_state(gmx_domdec_t *dd, t_state *state)
570 if (state->ddp_count != dd->ddp_count)
572 gmx_incons("The state does not the domain decomposition state");
575 state->ncg_gl = dd->ncg_home;
576 if (state->ncg_gl > state->cg_gl_nalloc)
578 state->cg_gl_nalloc = over_alloc_dd(state->ncg_gl);
579 srenew(state->cg_gl, state->cg_gl_nalloc);
581 for (i = 0; i < state->ncg_gl; i++)
583 state->cg_gl[i] = dd->index_gl[i];
586 state->ddp_count_cg_gl = dd->ddp_count;
589 gmx_domdec_zones_t *domdec_zones(gmx_domdec_t *dd)
591 return &dd->comm->zones;
594 void dd_get_ns_ranges(gmx_domdec_t *dd, int icg,
595 int *jcg0, int *jcg1, ivec shift0, ivec shift1)
597 gmx_domdec_zones_t *zones;
600 zones = &dd->comm->zones;
603 while (icg >= zones->izone[izone].cg1)
612 else if (izone < zones->nizone)
614 *jcg0 = zones->izone[izone].jcg0;
618 gmx_fatal(FARGS, "DD icg %d out of range: izone (%d) >= nizone (%d)",
619 icg, izone, zones->nizone);
622 *jcg1 = zones->izone[izone].jcg1;
624 for (d = 0; d < dd->ndim; d++)
627 shift0[dim] = zones->izone[izone].shift0[dim];
628 shift1[dim] = zones->izone[izone].shift1[dim];
629 if (dd->comm->tric_dir[dim] || (dd->bGridJump && d > 0))
631 /* A conservative approach, this can be optimized */
638 int dd_natoms_vsite(gmx_domdec_t *dd)
640 return dd->comm->nat[ddnatVSITE];
643 void dd_get_constraint_range(gmx_domdec_t *dd, int *at_start, int *at_end)
645 *at_start = dd->comm->nat[ddnatCON-1];
646 *at_end = dd->comm->nat[ddnatCON];
649 void dd_move_x(gmx_domdec_t *dd, matrix box, rvec x[])
651 int nzone, nat_tot, n, d, p, i, j, at0, at1, zone;
652 int *index, *cgindex;
653 gmx_domdec_comm_t *comm;
654 gmx_domdec_comm_dim_t *cd;
655 gmx_domdec_ind_t *ind;
656 rvec shift = {0, 0, 0}, *buf, *rbuf;
657 gmx_bool bPBC, bScrew;
661 cgindex = dd->cgindex;
666 nat_tot = dd->nat_home;
667 for (d = 0; d < dd->ndim; d++)
669 bPBC = (dd->ci[dd->dim[d]] == 0);
670 bScrew = (bPBC && dd->bScrewPBC && dd->dim[d] == XX);
673 copy_rvec(box[dd->dim[d]], shift);
676 for (p = 0; p < cd->np; p++)
683 for (i = 0; i < ind->nsend[nzone]; i++)
685 at0 = cgindex[index[i]];
686 at1 = cgindex[index[i]+1];
687 for (j = at0; j < at1; j++)
689 copy_rvec(x[j], buf[n]);
696 for (i = 0; i < ind->nsend[nzone]; i++)
698 at0 = cgindex[index[i]];
699 at1 = cgindex[index[i]+1];
700 for (j = at0; j < at1; j++)
702 /* We need to shift the coordinates */
703 rvec_add(x[j], shift, buf[n]);
710 for (i = 0; i < ind->nsend[nzone]; i++)
712 at0 = cgindex[index[i]];
713 at1 = cgindex[index[i]+1];
714 for (j = at0; j < at1; j++)
717 buf[n][XX] = x[j][XX] + shift[XX];
719 * This operation requires a special shift force
720 * treatment, which is performed in calc_vir.
722 buf[n][YY] = box[YY][YY] - x[j][YY];
723 buf[n][ZZ] = box[ZZ][ZZ] - x[j][ZZ];
735 rbuf = comm->vbuf2.v;
737 /* Send and receive the coordinates */
738 dd_sendrecv_rvec(dd, d, dddirBackward,
739 buf, ind->nsend[nzone+1],
740 rbuf, ind->nrecv[nzone+1]);
744 for (zone = 0; zone < nzone; zone++)
746 for (i = ind->cell2at0[zone]; i < ind->cell2at1[zone]; i++)
748 copy_rvec(rbuf[j], x[i]);
753 nat_tot += ind->nrecv[nzone+1];
759 void dd_move_f(gmx_domdec_t *dd, rvec f[], rvec *fshift)
761 int nzone, nat_tot, n, d, p, i, j, at0, at1, zone;
762 int *index, *cgindex;
763 gmx_domdec_comm_t *comm;
764 gmx_domdec_comm_dim_t *cd;
765 gmx_domdec_ind_t *ind;
769 gmx_bool bPBC, bScrew;
773 cgindex = dd->cgindex;
778 nzone = comm->zones.n/2;
779 nat_tot = dd->nat_tot;
780 for (d = dd->ndim-1; d >= 0; d--)
782 bPBC = (dd->ci[dd->dim[d]] == 0);
783 bScrew = (bPBC && dd->bScrewPBC && dd->dim[d] == XX);
784 if (fshift == NULL && !bScrew)
788 /* Determine which shift vector we need */
794 for (p = cd->np-1; p >= 0; p--)
797 nat_tot -= ind->nrecv[nzone+1];
804 sbuf = comm->vbuf2.v;
806 for (zone = 0; zone < nzone; zone++)
808 for (i = ind->cell2at0[zone]; i < ind->cell2at1[zone]; i++)
810 copy_rvec(f[i], sbuf[j]);
815 /* Communicate the forces */
816 dd_sendrecv_rvec(dd, d, dddirForward,
817 sbuf, ind->nrecv[nzone+1],
818 buf, ind->nsend[nzone+1]);
820 /* Add the received forces */
824 for (i = 0; i < ind->nsend[nzone]; i++)
826 at0 = cgindex[index[i]];
827 at1 = cgindex[index[i]+1];
828 for (j = at0; j < at1; j++)
830 rvec_inc(f[j], buf[n]);
837 for (i = 0; i < ind->nsend[nzone]; i++)
839 at0 = cgindex[index[i]];
840 at1 = cgindex[index[i]+1];
841 for (j = at0; j < at1; j++)
843 rvec_inc(f[j], buf[n]);
844 /* Add this force to the shift force */
845 rvec_inc(fshift[is], buf[n]);
852 for (i = 0; i < ind->nsend[nzone]; i++)
854 at0 = cgindex[index[i]];
855 at1 = cgindex[index[i]+1];
856 for (j = at0; j < at1; j++)
858 /* Rotate the force */
859 f[j][XX] += buf[n][XX];
860 f[j][YY] -= buf[n][YY];
861 f[j][ZZ] -= buf[n][ZZ];
864 /* Add this force to the shift force */
865 rvec_inc(fshift[is], buf[n]);
876 void dd_atom_spread_real(gmx_domdec_t *dd, real v[])
878 int nzone, nat_tot, n, d, p, i, j, at0, at1, zone;
879 int *index, *cgindex;
880 gmx_domdec_comm_t *comm;
881 gmx_domdec_comm_dim_t *cd;
882 gmx_domdec_ind_t *ind;
887 cgindex = dd->cgindex;
889 buf = &comm->vbuf.v[0][0];
892 nat_tot = dd->nat_home;
893 for (d = 0; d < dd->ndim; d++)
896 for (p = 0; p < cd->np; p++)
901 for (i = 0; i < ind->nsend[nzone]; i++)
903 at0 = cgindex[index[i]];
904 at1 = cgindex[index[i]+1];
905 for (j = at0; j < at1; j++)
918 rbuf = &comm->vbuf2.v[0][0];
920 /* Send and receive the coordinates */
921 dd_sendrecv_real(dd, d, dddirBackward,
922 buf, ind->nsend[nzone+1],
923 rbuf, ind->nrecv[nzone+1]);
927 for (zone = 0; zone < nzone; zone++)
929 for (i = ind->cell2at0[zone]; i < ind->cell2at1[zone]; i++)
936 nat_tot += ind->nrecv[nzone+1];
942 void dd_atom_sum_real(gmx_domdec_t *dd, real v[])
944 int nzone, nat_tot, n, d, p, i, j, at0, at1, zone;
945 int *index, *cgindex;
946 gmx_domdec_comm_t *comm;
947 gmx_domdec_comm_dim_t *cd;
948 gmx_domdec_ind_t *ind;
953 cgindex = dd->cgindex;
955 buf = &comm->vbuf.v[0][0];
958 nzone = comm->zones.n/2;
959 nat_tot = dd->nat_tot;
960 for (d = dd->ndim-1; d >= 0; d--)
963 for (p = cd->np-1; p >= 0; p--)
966 nat_tot -= ind->nrecv[nzone+1];
973 sbuf = &comm->vbuf2.v[0][0];
975 for (zone = 0; zone < nzone; zone++)
977 for (i = ind->cell2at0[zone]; i < ind->cell2at1[zone]; i++)
984 /* Communicate the forces */
985 dd_sendrecv_real(dd, d, dddirForward,
986 sbuf, ind->nrecv[nzone+1],
987 buf, ind->nsend[nzone+1]);
989 /* Add the received forces */
991 for (i = 0; i < ind->nsend[nzone]; i++)
993 at0 = cgindex[index[i]];
994 at1 = cgindex[index[i]+1];
995 for (j = at0; j < at1; j++)
1006 static void print_ddzone(FILE *fp, int d, int i, int j, gmx_ddzone_t *zone)
1008 fprintf(fp, "zone d0 %d d1 %d d2 %d min0 %6.3f max1 %6.3f mch0 %6.3f mch1 %6.3f p1_0 %6.3f p1_1 %6.3f\n",
1010 zone->min0, zone->max1,
1011 zone->mch0, zone->mch0,
1012 zone->p1_0, zone->p1_1);
1016 #define DDZONECOMM_MAXZONE 5
1017 #define DDZONECOMM_BUFSIZE 3
1019 static void dd_sendrecv_ddzone(const gmx_domdec_t *dd,
1020 int ddimind, int direction,
1021 gmx_ddzone_t *buf_s, int n_s,
1022 gmx_ddzone_t *buf_r, int n_r)
1024 #define ZBS DDZONECOMM_BUFSIZE
1025 rvec vbuf_s[DDZONECOMM_MAXZONE*ZBS];
1026 rvec vbuf_r[DDZONECOMM_MAXZONE*ZBS];
1029 for (i = 0; i < n_s; i++)
1031 vbuf_s[i*ZBS ][0] = buf_s[i].min0;
1032 vbuf_s[i*ZBS ][1] = buf_s[i].max1;
1033 vbuf_s[i*ZBS ][2] = buf_s[i].min1;
1034 vbuf_s[i*ZBS+1][0] = buf_s[i].mch0;
1035 vbuf_s[i*ZBS+1][1] = buf_s[i].mch1;
1036 vbuf_s[i*ZBS+1][2] = 0;
1037 vbuf_s[i*ZBS+2][0] = buf_s[i].p1_0;
1038 vbuf_s[i*ZBS+2][1] = buf_s[i].p1_1;
1039 vbuf_s[i*ZBS+2][2] = 0;
1042 dd_sendrecv_rvec(dd, ddimind, direction,
1046 for (i = 0; i < n_r; i++)
1048 buf_r[i].min0 = vbuf_r[i*ZBS ][0];
1049 buf_r[i].max1 = vbuf_r[i*ZBS ][1];
1050 buf_r[i].min1 = vbuf_r[i*ZBS ][2];
1051 buf_r[i].mch0 = vbuf_r[i*ZBS+1][0];
1052 buf_r[i].mch1 = vbuf_r[i*ZBS+1][1];
1053 buf_r[i].p1_0 = vbuf_r[i*ZBS+2][0];
1054 buf_r[i].p1_1 = vbuf_r[i*ZBS+2][1];
1060 static void dd_move_cellx(gmx_domdec_t *dd, gmx_ddbox_t *ddbox,
1061 rvec cell_ns_x0, rvec cell_ns_x1)
1063 int d, d1, dim, dim1, pos, buf_size, i, j, k, p, npulse, npulse_min;
1065 gmx_ddzone_t buf_s[DDZONECOMM_MAXZONE];
1066 gmx_ddzone_t buf_r[DDZONECOMM_MAXZONE];
1067 gmx_ddzone_t buf_e[DDZONECOMM_MAXZONE];
1068 rvec extr_s[2], extr_r[2];
1070 real dist_d, c = 0, det;
1071 gmx_domdec_comm_t *comm;
1072 gmx_bool bPBC, bUse;
1076 for (d = 1; d < dd->ndim; d++)
1079 zp = (d == 1) ? &comm->zone_d1[0] : &comm->zone_d2[0][0];
1080 zp->min0 = cell_ns_x0[dim];
1081 zp->max1 = cell_ns_x1[dim];
1082 zp->min1 = cell_ns_x1[dim];
1083 zp->mch0 = cell_ns_x0[dim];
1084 zp->mch1 = cell_ns_x1[dim];
1085 zp->p1_0 = cell_ns_x0[dim];
1086 zp->p1_1 = cell_ns_x1[dim];
1089 for (d = dd->ndim-2; d >= 0; d--)
1092 bPBC = (dim < ddbox->npbcdim);
1094 /* Use an rvec to store two reals */
1095 extr_s[d][0] = comm->cell_f0[d+1];
1096 extr_s[d][1] = comm->cell_f1[d+1];
1097 extr_s[d][2] = comm->cell_f1[d+1];
1100 /* Store the extremes in the backward sending buffer,
1101 * so the get updated separately from the forward communication.
1103 for (d1 = d; d1 < dd->ndim-1; d1++)
1105 /* We invert the order to be able to use the same loop for buf_e */
1106 buf_s[pos].min0 = extr_s[d1][1];
1107 buf_s[pos].max1 = extr_s[d1][0];
1108 buf_s[pos].min1 = extr_s[d1][2];
1109 buf_s[pos].mch0 = 0;
1110 buf_s[pos].mch1 = 0;
1111 /* Store the cell corner of the dimension we communicate along */
1112 buf_s[pos].p1_0 = comm->cell_x0[dim];
1113 buf_s[pos].p1_1 = 0;
1117 buf_s[pos] = (dd->ndim == 2) ? comm->zone_d1[0] : comm->zone_d2[0][0];
1120 if (dd->ndim == 3 && d == 0)
1122 buf_s[pos] = comm->zone_d2[0][1];
1124 buf_s[pos] = comm->zone_d1[0];
1128 /* We only need to communicate the extremes
1129 * in the forward direction
1131 npulse = comm->cd[d].np;
1134 /* Take the minimum to avoid double communication */
1135 npulse_min = min(npulse, dd->nc[dim]-1-npulse);
1139 /* Without PBC we should really not communicate over
1140 * the boundaries, but implementing that complicates
1141 * the communication setup and therefore we simply
1142 * do all communication, but ignore some data.
1144 npulse_min = npulse;
1146 for (p = 0; p < npulse_min; p++)
1148 /* Communicate the extremes forward */
1149 bUse = (bPBC || dd->ci[dim] > 0);
1151 dd_sendrecv_rvec(dd, d, dddirForward,
1152 extr_s+d, dd->ndim-d-1,
1153 extr_r+d, dd->ndim-d-1);
1157 for (d1 = d; d1 < dd->ndim-1; d1++)
1159 extr_s[d1][0] = max(extr_s[d1][0], extr_r[d1][0]);
1160 extr_s[d1][1] = min(extr_s[d1][1], extr_r[d1][1]);
1161 extr_s[d1][2] = min(extr_s[d1][2], extr_r[d1][2]);
1167 for (p = 0; p < npulse; p++)
1169 /* Communicate all the zone information backward */
1170 bUse = (bPBC || dd->ci[dim] < dd->nc[dim] - 1);
1172 dd_sendrecv_ddzone(dd, d, dddirBackward,
1179 for (d1 = d+1; d1 < dd->ndim; d1++)
1181 /* Determine the decrease of maximum required
1182 * communication height along d1 due to the distance along d,
1183 * this avoids a lot of useless atom communication.
1185 dist_d = comm->cell_x1[dim] - buf_r[0].p1_0;
1187 if (ddbox->tric_dir[dim])
1189 /* c is the off-diagonal coupling between the cell planes
1190 * along directions d and d1.
1192 c = ddbox->v[dim][dd->dim[d1]][dim];
1198 det = (1 + c*c)*comm->cutoff*comm->cutoff - dist_d*dist_d;
1201 dh[d1] = comm->cutoff - (c*dist_d + sqrt(det))/(1 + c*c);
1205 /* A negative value signals out of range */
1211 /* Accumulate the extremes over all pulses */
1212 for (i = 0; i < buf_size; i++)
1216 buf_e[i] = buf_r[i];
1222 buf_e[i].min0 = min(buf_e[i].min0, buf_r[i].min0);
1223 buf_e[i].max1 = max(buf_e[i].max1, buf_r[i].max1);
1224 buf_e[i].min1 = min(buf_e[i].min1, buf_r[i].min1);
1227 if (dd->ndim == 3 && d == 0 && i == buf_size - 1)
1235 if (bUse && dh[d1] >= 0)
1237 buf_e[i].mch0 = max(buf_e[i].mch0, buf_r[i].mch0-dh[d1]);
1238 buf_e[i].mch1 = max(buf_e[i].mch1, buf_r[i].mch1-dh[d1]);
1241 /* Copy the received buffer to the send buffer,
1242 * to pass the data through with the next pulse.
1244 buf_s[i] = buf_r[i];
1246 if (((bPBC || dd->ci[dim]+npulse < dd->nc[dim]) && p == npulse-1) ||
1247 (!bPBC && dd->ci[dim]+1+p == dd->nc[dim]-1))
1249 /* Store the extremes */
1252 for (d1 = d; d1 < dd->ndim-1; d1++)
1254 extr_s[d1][1] = min(extr_s[d1][1], buf_e[pos].min0);
1255 extr_s[d1][0] = max(extr_s[d1][0], buf_e[pos].max1);
1256 extr_s[d1][2] = min(extr_s[d1][2], buf_e[pos].min1);
1260 if (d == 1 || (d == 0 && dd->ndim == 3))
1262 for (i = d; i < 2; i++)
1264 comm->zone_d2[1-d][i] = buf_e[pos];
1270 comm->zone_d1[1] = buf_e[pos];
1280 for (i = 0; i < 2; i++)
1284 print_ddzone(debug, 1, i, 0, &comm->zone_d1[i]);
1286 cell_ns_x0[dim] = min(cell_ns_x0[dim], comm->zone_d1[i].min0);
1287 cell_ns_x1[dim] = max(cell_ns_x1[dim], comm->zone_d1[i].max1);
1293 for (i = 0; i < 2; i++)
1295 for (j = 0; j < 2; j++)
1299 print_ddzone(debug, 2, i, j, &comm->zone_d2[i][j]);
1301 cell_ns_x0[dim] = min(cell_ns_x0[dim], comm->zone_d2[i][j].min0);
1302 cell_ns_x1[dim] = max(cell_ns_x1[dim], comm->zone_d2[i][j].max1);
1306 for (d = 1; d < dd->ndim; d++)
1308 comm->cell_f_max0[d] = extr_s[d-1][0];
1309 comm->cell_f_min1[d] = extr_s[d-1][1];
1312 fprintf(debug, "Cell fraction d %d, max0 %f, min1 %f\n",
1313 d, comm->cell_f_max0[d], comm->cell_f_min1[d]);
1318 static void dd_collect_cg(gmx_domdec_t *dd,
1319 t_state *state_local)
1321 gmx_domdec_master_t *ma = NULL;
1322 int buf2[2], *ibuf, i, ncg_home = 0, *cg = NULL, nat_home = 0;
1324 if (state_local->ddp_count == dd->comm->master_cg_ddp_count)
1326 /* The master has the correct distribution */
1330 if (state_local->ddp_count == dd->ddp_count)
1332 /* The local state and DD are in sync, use the DD indices */
1333 ncg_home = dd->ncg_home;
1335 nat_home = dd->nat_home;
1337 else if (state_local->ddp_count_cg_gl == state_local->ddp_count)
1339 /* The DD is out of sync with the local state, but we have stored
1340 * the cg indices with the local state, so we can use those.
1344 cgs_gl = &dd->comm->cgs_gl;
1346 ncg_home = state_local->ncg_gl;
1347 cg = state_local->cg_gl;
1349 for (i = 0; i < ncg_home; i++)
1351 nat_home += cgs_gl->index[cg[i]+1] - cgs_gl->index[cg[i]];
1356 gmx_incons("Attempted to collect a vector for a state for which the charge group distribution is unknown");
1370 /* Collect the charge group and atom counts on the master */
1371 dd_gather(dd, 2*sizeof(int), buf2, ibuf);
1376 for (i = 0; i < dd->nnodes; i++)
1378 ma->ncg[i] = ma->ibuf[2*i];
1379 ma->nat[i] = ma->ibuf[2*i+1];
1380 ma->index[i+1] = ma->index[i] + ma->ncg[i];
1383 /* Make byte counts and indices */
1384 for (i = 0; i < dd->nnodes; i++)
1386 ma->ibuf[i] = ma->ncg[i]*sizeof(int);
1387 ma->ibuf[dd->nnodes+i] = ma->index[i]*sizeof(int);
1391 fprintf(debug, "Initial charge group distribution: ");
1392 for (i = 0; i < dd->nnodes; i++)
1394 fprintf(debug, " %d", ma->ncg[i]);
1396 fprintf(debug, "\n");
1400 /* Collect the charge group indices on the master */
1402 ncg_home*sizeof(int), cg,
1403 DDMASTER(dd) ? ma->ibuf : NULL,
1404 DDMASTER(dd) ? ma->ibuf+dd->nnodes : NULL,
1405 DDMASTER(dd) ? ma->cg : NULL);
1407 dd->comm->master_cg_ddp_count = state_local->ddp_count;
1410 static void dd_collect_vec_sendrecv(gmx_domdec_t *dd,
1413 gmx_domdec_master_t *ma;
1414 int n, i, c, a, nalloc = 0;
1423 MPI_Send(lv, dd->nat_home*sizeof(rvec), MPI_BYTE, DDMASTERRANK(dd),
1424 dd->rank, dd->mpi_comm_all);
1429 /* Copy the master coordinates to the global array */
1430 cgs_gl = &dd->comm->cgs_gl;
1432 n = DDMASTERRANK(dd);
1434 for (i = ma->index[n]; i < ma->index[n+1]; i++)
1436 for (c = cgs_gl->index[ma->cg[i]]; c < cgs_gl->index[ma->cg[i]+1]; c++)
1438 copy_rvec(lv[a++], v[c]);
1442 for (n = 0; n < dd->nnodes; n++)
1446 if (ma->nat[n] > nalloc)
1448 nalloc = over_alloc_dd(ma->nat[n]);
1449 srenew(buf, nalloc);
1452 MPI_Recv(buf, ma->nat[n]*sizeof(rvec), MPI_BYTE, DDRANK(dd, n),
1453 n, dd->mpi_comm_all, MPI_STATUS_IGNORE);
1456 for (i = ma->index[n]; i < ma->index[n+1]; i++)
1458 for (c = cgs_gl->index[ma->cg[i]]; c < cgs_gl->index[ma->cg[i]+1]; c++)
1460 copy_rvec(buf[a++], v[c]);
1469 static void get_commbuffer_counts(gmx_domdec_t *dd,
1470 int **counts, int **disps)
1472 gmx_domdec_master_t *ma;
1477 /* Make the rvec count and displacment arrays */
1479 *disps = ma->ibuf + dd->nnodes;
1480 for (n = 0; n < dd->nnodes; n++)
1482 (*counts)[n] = ma->nat[n]*sizeof(rvec);
1483 (*disps)[n] = (n == 0 ? 0 : (*disps)[n-1] + (*counts)[n-1]);
1487 static void dd_collect_vec_gatherv(gmx_domdec_t *dd,
1490 gmx_domdec_master_t *ma;
1491 int *rcounts = NULL, *disps = NULL;
1500 get_commbuffer_counts(dd, &rcounts, &disps);
1505 dd_gatherv(dd, dd->nat_home*sizeof(rvec), lv, rcounts, disps, buf);
1509 cgs_gl = &dd->comm->cgs_gl;
1512 for (n = 0; n < dd->nnodes; n++)
1514 for (i = ma->index[n]; i < ma->index[n+1]; i++)
1516 for (c = cgs_gl->index[ma->cg[i]]; c < cgs_gl->index[ma->cg[i]+1]; c++)
1518 copy_rvec(buf[a++], v[c]);
1525 void dd_collect_vec(gmx_domdec_t *dd,
1526 t_state *state_local, rvec *lv, rvec *v)
1528 gmx_domdec_master_t *ma;
1529 int n, i, c, a, nalloc = 0;
1532 dd_collect_cg(dd, state_local);
1534 if (dd->nnodes <= GMX_DD_NNODES_SENDRECV)
1536 dd_collect_vec_sendrecv(dd, lv, v);
1540 dd_collect_vec_gatherv(dd, lv, v);
1545 void dd_collect_state(gmx_domdec_t *dd,
1546 t_state *state_local, t_state *state)
1550 nh = state->nhchainlength;
1554 for (i = 0; i < efptNR; i++)
1556 state->lambda[i] = state_local->lambda[i];
1558 state->fep_state = state_local->fep_state;
1559 state->veta = state_local->veta;
1560 state->vol0 = state_local->vol0;
1561 copy_mat(state_local->box, state->box);
1562 copy_mat(state_local->boxv, state->boxv);
1563 copy_mat(state_local->svir_prev, state->svir_prev);
1564 copy_mat(state_local->fvir_prev, state->fvir_prev);
1565 copy_mat(state_local->pres_prev, state->pres_prev);
1567 for (i = 0; i < state_local->ngtc; i++)
1569 for (j = 0; j < nh; j++)
1571 state->nosehoover_xi[i*nh+j] = state_local->nosehoover_xi[i*nh+j];
1572 state->nosehoover_vxi[i*nh+j] = state_local->nosehoover_vxi[i*nh+j];
1574 state->therm_integral[i] = state_local->therm_integral[i];
1576 for (i = 0; i < state_local->nnhpres; i++)
1578 for (j = 0; j < nh; j++)
1580 state->nhpres_xi[i*nh+j] = state_local->nhpres_xi[i*nh+j];
1581 state->nhpres_vxi[i*nh+j] = state_local->nhpres_vxi[i*nh+j];
1585 for (est = 0; est < estNR; est++)
1587 if (EST_DISTR(est) && (state_local->flags & (1<<est)))
1592 dd_collect_vec(dd, state_local, state_local->x, state->x);
1595 dd_collect_vec(dd, state_local, state_local->v, state->v);
1598 dd_collect_vec(dd, state_local, state_local->sd_X, state->sd_X);
1601 dd_collect_vec(dd, state_local, state_local->cg_p, state->cg_p);
1603 case estDISRE_INITF:
1604 case estDISRE_RM3TAV:
1605 case estORIRE_INITF:
1609 gmx_incons("Unknown state entry encountered in dd_collect_state");
1615 static void dd_realloc_state(t_state *state, rvec **f, int nalloc)
1621 fprintf(debug, "Reallocating state: currently %d, required %d, allocating %d\n", state->nalloc, nalloc, over_alloc_dd(nalloc));
1624 state->nalloc = over_alloc_dd(nalloc);
1626 for (est = 0; est < estNR; est++)
1628 if (EST_DISTR(est) && (state->flags & (1<<est)))
1633 srenew(state->x, state->nalloc);
1636 srenew(state->v, state->nalloc);
1639 srenew(state->sd_X, state->nalloc);
1642 srenew(state->cg_p, state->nalloc);
1644 case estDISRE_INITF:
1645 case estDISRE_RM3TAV:
1646 case estORIRE_INITF:
1648 /* No reallocation required */
1651 gmx_incons("Unknown state entry encountered in dd_realloc_state");
1658 srenew(*f, state->nalloc);
1662 static void dd_check_alloc_ncg(t_forcerec *fr, t_state *state, rvec **f,
1665 if (nalloc > fr->cg_nalloc)
1669 fprintf(debug, "Reallocating forcerec: currently %d, required %d, allocating %d\n", fr->cg_nalloc, nalloc, over_alloc_dd(nalloc));
1671 fr->cg_nalloc = over_alloc_dd(nalloc);
1672 srenew(fr->cginfo, fr->cg_nalloc);
1673 if (fr->cutoff_scheme == ecutsGROUP)
1675 srenew(fr->cg_cm, fr->cg_nalloc);
1678 if (fr->cutoff_scheme == ecutsVERLET && nalloc > state->nalloc)
1680 /* We don't use charge groups, we use x in state to set up
1681 * the atom communication.
1683 dd_realloc_state(state, f, nalloc);
1687 static void dd_distribute_vec_sendrecv(gmx_domdec_t *dd, t_block *cgs,
1690 gmx_domdec_master_t *ma;
1691 int n, i, c, a, nalloc = 0;
1698 for (n = 0; n < dd->nnodes; n++)
1702 if (ma->nat[n] > nalloc)
1704 nalloc = over_alloc_dd(ma->nat[n]);
1705 srenew(buf, nalloc);
1707 /* Use lv as a temporary buffer */
1709 for (i = ma->index[n]; i < ma->index[n+1]; i++)
1711 for (c = cgs->index[ma->cg[i]]; c < cgs->index[ma->cg[i]+1]; c++)
1713 copy_rvec(v[c], buf[a++]);
1716 if (a != ma->nat[n])
1718 gmx_fatal(FARGS, "Internal error a (%d) != nat (%d)",
1723 MPI_Send(buf, ma->nat[n]*sizeof(rvec), MPI_BYTE,
1724 DDRANK(dd, n), n, dd->mpi_comm_all);
1729 n = DDMASTERRANK(dd);
1731 for (i = ma->index[n]; i < ma->index[n+1]; i++)
1733 for (c = cgs->index[ma->cg[i]]; c < cgs->index[ma->cg[i]+1]; c++)
1735 copy_rvec(v[c], lv[a++]);
1742 MPI_Recv(lv, dd->nat_home*sizeof(rvec), MPI_BYTE, DDMASTERRANK(dd),
1743 MPI_ANY_TAG, dd->mpi_comm_all, MPI_STATUS_IGNORE);
1748 static void dd_distribute_vec_scatterv(gmx_domdec_t *dd, t_block *cgs,
1751 gmx_domdec_master_t *ma;
1752 int *scounts = NULL, *disps = NULL;
1753 int n, i, c, a, nalloc = 0;
1760 get_commbuffer_counts(dd, &scounts, &disps);
1764 for (n = 0; n < dd->nnodes; n++)
1766 for (i = ma->index[n]; i < ma->index[n+1]; i++)
1768 for (c = cgs->index[ma->cg[i]]; c < cgs->index[ma->cg[i]+1]; c++)
1770 copy_rvec(v[c], buf[a++]);
1776 dd_scatterv(dd, scounts, disps, buf, dd->nat_home*sizeof(rvec), lv);
1779 static void dd_distribute_vec(gmx_domdec_t *dd, t_block *cgs, rvec *v, rvec *lv)
1781 if (dd->nnodes <= GMX_DD_NNODES_SENDRECV)
1783 dd_distribute_vec_sendrecv(dd, cgs, v, lv);
1787 dd_distribute_vec_scatterv(dd, cgs, v, lv);
1791 static void dd_distribute_dfhist(gmx_domdec_t *dd, df_history_t *dfhist)
1794 dd_bcast(dd, sizeof(int), &dfhist->bEquil);
1795 dd_bcast(dd, sizeof(int), &dfhist->nlambda);
1796 dd_bcast(dd, sizeof(real), &dfhist->wl_delta);
1798 if (dfhist->nlambda > 0)
1800 int nlam = dfhist->nlambda;
1801 dd_bcast(dd, sizeof(int)*nlam, dfhist->n_at_lam);
1802 dd_bcast(dd, sizeof(real)*nlam, dfhist->wl_histo);
1803 dd_bcast(dd, sizeof(real)*nlam, dfhist->sum_weights);
1804 dd_bcast(dd, sizeof(real)*nlam, dfhist->sum_dg);
1805 dd_bcast(dd, sizeof(real)*nlam, dfhist->sum_minvar);
1806 dd_bcast(dd, sizeof(real)*nlam, dfhist->sum_variance);
1808 for (i = 0; i < nlam; i++)
1810 dd_bcast(dd, sizeof(real)*nlam, dfhist->accum_p[i]);
1811 dd_bcast(dd, sizeof(real)*nlam, dfhist->accum_m[i]);
1812 dd_bcast(dd, sizeof(real)*nlam, dfhist->accum_p2[i]);
1813 dd_bcast(dd, sizeof(real)*nlam, dfhist->accum_m2[i]);
1814 dd_bcast(dd, sizeof(real)*nlam, dfhist->Tij[i]);
1815 dd_bcast(dd, sizeof(real)*nlam, dfhist->Tij_empirical[i]);
1820 static void dd_distribute_state(gmx_domdec_t *dd, t_block *cgs,
1821 t_state *state, t_state *state_local,
1826 nh = state->nhchainlength;
1830 for (i = 0; i < efptNR; i++)
1832 state_local->lambda[i] = state->lambda[i];
1834 state_local->fep_state = state->fep_state;
1835 state_local->veta = state->veta;
1836 state_local->vol0 = state->vol0;
1837 copy_mat(state->box, state_local->box);
1838 copy_mat(state->box_rel, state_local->box_rel);
1839 copy_mat(state->boxv, state_local->boxv);
1840 copy_mat(state->svir_prev, state_local->svir_prev);
1841 copy_mat(state->fvir_prev, state_local->fvir_prev);
1842 copy_df_history(&state_local->dfhist, &state->dfhist);
1843 for (i = 0; i < state_local->ngtc; i++)
1845 for (j = 0; j < nh; j++)
1847 state_local->nosehoover_xi[i*nh+j] = state->nosehoover_xi[i*nh+j];
1848 state_local->nosehoover_vxi[i*nh+j] = state->nosehoover_vxi[i*nh+j];
1850 state_local->therm_integral[i] = state->therm_integral[i];
1852 for (i = 0; i < state_local->nnhpres; i++)
1854 for (j = 0; j < nh; j++)
1856 state_local->nhpres_xi[i*nh+j] = state->nhpres_xi[i*nh+j];
1857 state_local->nhpres_vxi[i*nh+j] = state->nhpres_vxi[i*nh+j];
1861 dd_bcast(dd, ((efptNR)*sizeof(real)), state_local->lambda);
1862 dd_bcast(dd, sizeof(int), &state_local->fep_state);
1863 dd_bcast(dd, sizeof(real), &state_local->veta);
1864 dd_bcast(dd, sizeof(real), &state_local->vol0);
1865 dd_bcast(dd, sizeof(state_local->box), state_local->box);
1866 dd_bcast(dd, sizeof(state_local->box_rel), state_local->box_rel);
1867 dd_bcast(dd, sizeof(state_local->boxv), state_local->boxv);
1868 dd_bcast(dd, sizeof(state_local->svir_prev), state_local->svir_prev);
1869 dd_bcast(dd, sizeof(state_local->fvir_prev), state_local->fvir_prev);
1870 dd_bcast(dd, ((state_local->ngtc*nh)*sizeof(double)), state_local->nosehoover_xi);
1871 dd_bcast(dd, ((state_local->ngtc*nh)*sizeof(double)), state_local->nosehoover_vxi);
1872 dd_bcast(dd, state_local->ngtc*sizeof(double), state_local->therm_integral);
1873 dd_bcast(dd, ((state_local->nnhpres*nh)*sizeof(double)), state_local->nhpres_xi);
1874 dd_bcast(dd, ((state_local->nnhpres*nh)*sizeof(double)), state_local->nhpres_vxi);
1876 /* communicate df_history -- required for restarting from checkpoint */
1877 dd_distribute_dfhist(dd, &state_local->dfhist);
1879 if (dd->nat_home > state_local->nalloc)
1881 dd_realloc_state(state_local, f, dd->nat_home);
1883 for (i = 0; i < estNR; i++)
1885 if (EST_DISTR(i) && (state_local->flags & (1<<i)))
1890 dd_distribute_vec(dd, cgs, state->x, state_local->x);
1893 dd_distribute_vec(dd, cgs, state->v, state_local->v);
1896 dd_distribute_vec(dd, cgs, state->sd_X, state_local->sd_X);
1899 dd_distribute_vec(dd, cgs, state->cg_p, state_local->cg_p);
1901 case estDISRE_INITF:
1902 case estDISRE_RM3TAV:
1903 case estORIRE_INITF:
1905 /* Not implemented yet */
1908 gmx_incons("Unknown state entry encountered in dd_distribute_state");
1914 static char dim2char(int dim)
1920 case XX: c = 'X'; break;
1921 case YY: c = 'Y'; break;
1922 case ZZ: c = 'Z'; break;
1923 default: gmx_fatal(FARGS, "Unknown dim %d", dim);
1929 static void write_dd_grid_pdb(const char *fn, gmx_int64_t step,
1930 gmx_domdec_t *dd, matrix box, gmx_ddbox_t *ddbox)
1932 rvec grid_s[2], *grid_r = NULL, cx, r;
1933 char fname[STRLEN], buf[22];
1935 int a, i, d, z, y, x;
1939 copy_rvec(dd->comm->cell_x0, grid_s[0]);
1940 copy_rvec(dd->comm->cell_x1, grid_s[1]);
1944 snew(grid_r, 2*dd->nnodes);
1947 dd_gather(dd, 2*sizeof(rvec), grid_s[0], DDMASTER(dd) ? grid_r[0] : NULL);
1951 for (d = 0; d < DIM; d++)
1953 for (i = 0; i < DIM; i++)
1961 if (d < ddbox->npbcdim && dd->nc[d] > 1)
1963 tric[d][i] = box[i][d]/box[i][i];
1972 sprintf(fname, "%s_%s.pdb", fn, gmx_step_str(step, buf));
1973 out = gmx_fio_fopen(fname, "w");
1974 gmx_write_pdb_box(out, dd->bScrewPBC ? epbcSCREW : epbcXYZ, box);
1976 for (i = 0; i < dd->nnodes; i++)
1978 vol = dd->nnodes/(box[XX][XX]*box[YY][YY]*box[ZZ][ZZ]);
1979 for (d = 0; d < DIM; d++)
1981 vol *= grid_r[i*2+1][d] - grid_r[i*2][d];
1983 for (z = 0; z < 2; z++)
1985 for (y = 0; y < 2; y++)
1987 for (x = 0; x < 2; x++)
1989 cx[XX] = grid_r[i*2+x][XX];
1990 cx[YY] = grid_r[i*2+y][YY];
1991 cx[ZZ] = grid_r[i*2+z][ZZ];
1993 gmx_fprintf_pdb_atomline(out, epdbATOM, a++, "CA", ' ', "GLY", ' ', i+1, ' ',
1994 10*r[XX], 10*r[YY], 10*r[ZZ], 1.0, vol, "");
1998 for (d = 0; d < DIM; d++)
2000 for (x = 0; x < 4; x++)
2004 case 0: y = 1 + i*8 + 2*x; break;
2005 case 1: y = 1 + i*8 + 2*x - (x % 2); break;
2006 case 2: y = 1 + i*8 + x; break;
2008 fprintf(out, "%6s%5d%5d\n", "CONECT", y, y+(1<<d));
2012 gmx_fio_fclose(out);
2017 void write_dd_pdb(const char *fn, gmx_int64_t step, const char *title,
2018 gmx_mtop_t *mtop, t_commrec *cr,
2019 int natoms, rvec x[], matrix box)
2021 char fname[STRLEN], buf[22];
2023 int i, ii, resnr, c;
2024 char *atomname, *resname;
2031 natoms = dd->comm->nat[ddnatVSITE];
2034 sprintf(fname, "%s_%s_n%d.pdb", fn, gmx_step_str(step, buf), cr->sim_nodeid);
2036 out = gmx_fio_fopen(fname, "w");
2038 fprintf(out, "TITLE %s\n", title);
2039 gmx_write_pdb_box(out, dd->bScrewPBC ? epbcSCREW : epbcXYZ, box);
2040 for (i = 0; i < natoms; i++)
2042 ii = dd->gatindex[i];
2043 gmx_mtop_atominfo_global(mtop, ii, &atomname, &resnr, &resname);
2044 if (i < dd->comm->nat[ddnatZONE])
2047 while (i >= dd->cgindex[dd->comm->zones.cg_range[c+1]])
2053 else if (i < dd->comm->nat[ddnatVSITE])
2055 b = dd->comm->zones.n;
2059 b = dd->comm->zones.n + 1;
2061 gmx_fprintf_pdb_atomline(out, epdbATOM, ii+1, atomname, ' ', resname, ' ', resnr, ' ',
2062 10*x[i][XX], 10*x[i][YY], 10*x[i][ZZ], 1.0, b, "");
2064 fprintf(out, "TER\n");
2066 gmx_fio_fclose(out);
2069 real dd_cutoff_mbody(gmx_domdec_t *dd)
2071 gmx_domdec_comm_t *comm;
2078 if (comm->bInterCGBondeds)
2080 if (comm->cutoff_mbody > 0)
2082 r = comm->cutoff_mbody;
2086 /* cutoff_mbody=0 means we do not have DLB */
2087 r = comm->cellsize_min[dd->dim[0]];
2088 for (di = 1; di < dd->ndim; di++)
2090 r = min(r, comm->cellsize_min[dd->dim[di]]);
2092 if (comm->bBondComm)
2094 r = max(r, comm->cutoff_mbody);
2098 r = min(r, comm->cutoff);
2106 real dd_cutoff_twobody(gmx_domdec_t *dd)
2110 r_mb = dd_cutoff_mbody(dd);
2112 return max(dd->comm->cutoff, r_mb);
2116 static void dd_cart_coord2pmecoord(gmx_domdec_t *dd, ivec coord, ivec coord_pme)
2120 nc = dd->nc[dd->comm->cartpmedim];
2121 ntot = dd->comm->ntot[dd->comm->cartpmedim];
2122 copy_ivec(coord, coord_pme);
2123 coord_pme[dd->comm->cartpmedim] =
2124 nc + (coord[dd->comm->cartpmedim]*(ntot - nc) + (ntot - nc)/2)/nc;
2127 static int low_ddindex2pmeindex(int ndd, int npme, int ddindex)
2129 /* Here we assign a PME node to communicate with this DD node
2130 * by assuming that the major index of both is x.
2131 * We add cr->npmenodes/2 to obtain an even distribution.
2133 return (ddindex*npme + npme/2)/ndd;
2136 static int ddindex2pmeindex(const gmx_domdec_t *dd, int ddindex)
2138 return low_ddindex2pmeindex(dd->nnodes, dd->comm->npmenodes, ddindex);
2141 static int cr_ddindex2pmeindex(const t_commrec *cr, int ddindex)
2143 return low_ddindex2pmeindex(cr->dd->nnodes, cr->npmenodes, ddindex);
2146 static int *dd_pmenodes(t_commrec *cr)
2151 snew(pmenodes, cr->npmenodes);
2153 for (i = 0; i < cr->dd->nnodes; i++)
2155 p0 = cr_ddindex2pmeindex(cr, i);
2156 p1 = cr_ddindex2pmeindex(cr, i+1);
2157 if (i+1 == cr->dd->nnodes || p1 > p0)
2161 fprintf(debug, "pmenode[%d] = %d\n", n, i+1+n);
2163 pmenodes[n] = i + 1 + n;
2171 static int gmx_ddcoord2pmeindex(t_commrec *cr, int x, int y, int z)
2174 ivec coords, coords_pme, nc;
2179 if (dd->comm->bCartesian) {
2180 gmx_ddindex2xyz(dd->nc,ddindex,coords);
2181 dd_coords2pmecoords(dd,coords,coords_pme);
2182 copy_ivec(dd->ntot,nc);
2183 nc[dd->cartpmedim] -= dd->nc[dd->cartpmedim];
2184 coords_pme[dd->cartpmedim] -= dd->nc[dd->cartpmedim];
2186 slab = (coords_pme[XX]*nc[YY] + coords_pme[YY])*nc[ZZ] + coords_pme[ZZ];
2188 slab = (ddindex*cr->npmenodes + cr->npmenodes/2)/dd->nnodes;
2194 slab = ddindex2pmeindex(dd, dd_index(dd->nc, coords));
2199 static int ddcoord2simnodeid(t_commrec *cr, int x, int y, int z)
2201 gmx_domdec_comm_t *comm;
2203 int ddindex, nodeid = -1;
2205 comm = cr->dd->comm;
2210 if (comm->bCartesianPP_PME)
2213 MPI_Cart_rank(cr->mpi_comm_mysim, coords, &nodeid);
2218 ddindex = dd_index(cr->dd->nc, coords);
2219 if (comm->bCartesianPP)
2221 nodeid = comm->ddindex2simnodeid[ddindex];
2227 nodeid = ddindex + gmx_ddcoord2pmeindex(cr, x, y, z);
2239 static int dd_simnode2pmenode(t_commrec *cr, int sim_nodeid)
2242 gmx_domdec_comm_t *comm;
2243 ivec coord, coord_pme;
2250 /* This assumes a uniform x domain decomposition grid cell size */
2251 if (comm->bCartesianPP_PME)
2254 MPI_Cart_coords(cr->mpi_comm_mysim, sim_nodeid, DIM, coord);
2255 if (coord[comm->cartpmedim] < dd->nc[comm->cartpmedim])
2257 /* This is a PP node */
2258 dd_cart_coord2pmecoord(dd, coord, coord_pme);
2259 MPI_Cart_rank(cr->mpi_comm_mysim, coord_pme, &pmenode);
2263 else if (comm->bCartesianPP)
2265 if (sim_nodeid < dd->nnodes)
2267 pmenode = dd->nnodes + ddindex2pmeindex(dd, sim_nodeid);
2272 /* This assumes DD cells with identical x coordinates
2273 * are numbered sequentially.
2275 if (dd->comm->pmenodes == NULL)
2277 if (sim_nodeid < dd->nnodes)
2279 /* The DD index equals the nodeid */
2280 pmenode = dd->nnodes + ddindex2pmeindex(dd, sim_nodeid);
2286 while (sim_nodeid > dd->comm->pmenodes[i])
2290 if (sim_nodeid < dd->comm->pmenodes[i])
2292 pmenode = dd->comm->pmenodes[i];
2300 void get_pme_nnodes(const gmx_domdec_t *dd,
2301 int *npmenodes_x, int *npmenodes_y)
2305 *npmenodes_x = dd->comm->npmenodes_x;
2306 *npmenodes_y = dd->comm->npmenodes_y;
2315 gmx_bool gmx_pmeonlynode(t_commrec *cr, int sim_nodeid)
2317 gmx_bool bPMEOnlyNode;
2319 if (DOMAINDECOMP(cr))
2321 bPMEOnlyNode = (dd_simnode2pmenode(cr, sim_nodeid) == -1);
2325 bPMEOnlyNode = FALSE;
2328 return bPMEOnlyNode;
2331 void get_pme_ddnodes(t_commrec *cr, int pmenodeid,
2332 int *nmy_ddnodes, int **my_ddnodes, int *node_peer)
2336 ivec coord, coord_pme;
2340 snew(*my_ddnodes, (dd->nnodes+cr->npmenodes-1)/cr->npmenodes);
2343 for (x = 0; x < dd->nc[XX]; x++)
2345 for (y = 0; y < dd->nc[YY]; y++)
2347 for (z = 0; z < dd->nc[ZZ]; z++)
2349 if (dd->comm->bCartesianPP_PME)
2354 dd_cart_coord2pmecoord(dd, coord, coord_pme);
2355 if (dd->ci[XX] == coord_pme[XX] &&
2356 dd->ci[YY] == coord_pme[YY] &&
2357 dd->ci[ZZ] == coord_pme[ZZ])
2359 (*my_ddnodes)[(*nmy_ddnodes)++] = ddcoord2simnodeid(cr, x, y, z);
2364 /* The slab corresponds to the nodeid in the PME group */
2365 if (gmx_ddcoord2pmeindex(cr, x, y, z) == pmenodeid)
2367 (*my_ddnodes)[(*nmy_ddnodes)++] = ddcoord2simnodeid(cr, x, y, z);
2374 /* The last PP-only node is the peer node */
2375 *node_peer = (*my_ddnodes)[*nmy_ddnodes-1];
2379 fprintf(debug, "Receive coordinates from PP ranks:");
2380 for (x = 0; x < *nmy_ddnodes; x++)
2382 fprintf(debug, " %d", (*my_ddnodes)[x]);
2384 fprintf(debug, "\n");
2388 static gmx_bool receive_vir_ener(t_commrec *cr)
2390 gmx_domdec_comm_t *comm;
2391 int pmenode, coords[DIM], rank;
2395 if (cr->npmenodes < cr->dd->nnodes)
2397 comm = cr->dd->comm;
2398 if (comm->bCartesianPP_PME)
2400 pmenode = dd_simnode2pmenode(cr, cr->sim_nodeid);
2402 MPI_Cart_coords(cr->mpi_comm_mysim, cr->sim_nodeid, DIM, coords);
2403 coords[comm->cartpmedim]++;
2404 if (coords[comm->cartpmedim] < cr->dd->nc[comm->cartpmedim])
2406 MPI_Cart_rank(cr->mpi_comm_mysim, coords, &rank);
2407 if (dd_simnode2pmenode(cr, rank) == pmenode)
2409 /* This is not the last PP node for pmenode */
2417 pmenode = dd_simnode2pmenode(cr, cr->sim_nodeid);
2418 if (cr->sim_nodeid+1 < cr->nnodes &&
2419 dd_simnode2pmenode(cr, cr->sim_nodeid+1) == pmenode)
2421 /* This is not the last PP node for pmenode */
2430 static void set_zones_ncg_home(gmx_domdec_t *dd)
2432 gmx_domdec_zones_t *zones;
2435 zones = &dd->comm->zones;
2437 zones->cg_range[0] = 0;
2438 for (i = 1; i < zones->n+1; i++)
2440 zones->cg_range[i] = dd->ncg_home;
2442 /* zone_ncg1[0] should always be equal to ncg_home */
2443 dd->comm->zone_ncg1[0] = dd->ncg_home;
2446 static void rebuild_cgindex(gmx_domdec_t *dd,
2447 const int *gcgs_index, t_state *state)
2449 int nat, i, *ind, *dd_cg_gl, *cgindex, cg_gl;
2452 dd_cg_gl = dd->index_gl;
2453 cgindex = dd->cgindex;
2456 for (i = 0; i < state->ncg_gl; i++)
2460 dd_cg_gl[i] = cg_gl;
2461 nat += gcgs_index[cg_gl+1] - gcgs_index[cg_gl];
2465 dd->ncg_home = state->ncg_gl;
2468 set_zones_ncg_home(dd);
2471 static int ddcginfo(const cginfo_mb_t *cginfo_mb, int cg)
2473 while (cg >= cginfo_mb->cg_end)
2478 return cginfo_mb->cginfo[(cg - cginfo_mb->cg_start) % cginfo_mb->cg_mod];
2481 static void dd_set_cginfo(int *index_gl, int cg0, int cg1,
2482 t_forcerec *fr, char *bLocalCG)
2484 cginfo_mb_t *cginfo_mb;
2490 cginfo_mb = fr->cginfo_mb;
2491 cginfo = fr->cginfo;
2493 for (cg = cg0; cg < cg1; cg++)
2495 cginfo[cg] = ddcginfo(cginfo_mb, index_gl[cg]);
2499 if (bLocalCG != NULL)
2501 for (cg = cg0; cg < cg1; cg++)
2503 bLocalCG[index_gl[cg]] = TRUE;
2508 static void make_dd_indices(gmx_domdec_t *dd,
2509 const int *gcgs_index, int cg_start)
2511 int nzone, zone, zone1, cg0, cg1, cg1_p1, cg, cg_gl, a, a_gl;
2512 int *zone2cg, *zone_ncg1, *index_gl, *gatindex;
2517 bLocalCG = dd->comm->bLocalCG;
2519 if (dd->nat_tot > dd->gatindex_nalloc)
2521 dd->gatindex_nalloc = over_alloc_dd(dd->nat_tot);
2522 srenew(dd->gatindex, dd->gatindex_nalloc);
2525 nzone = dd->comm->zones.n;
2526 zone2cg = dd->comm->zones.cg_range;
2527 zone_ncg1 = dd->comm->zone_ncg1;
2528 index_gl = dd->index_gl;
2529 gatindex = dd->gatindex;
2530 bCGs = dd->comm->bCGs;
2532 if (zone2cg[1] != dd->ncg_home)
2534 gmx_incons("dd->ncg_zone is not up to date");
2537 /* Make the local to global and global to local atom index */
2538 a = dd->cgindex[cg_start];
2539 for (zone = 0; zone < nzone; zone++)
2547 cg0 = zone2cg[zone];
2549 cg1 = zone2cg[zone+1];
2550 cg1_p1 = cg0 + zone_ncg1[zone];
2552 for (cg = cg0; cg < cg1; cg++)
2557 /* Signal that this cg is from more than one pulse away */
2560 cg_gl = index_gl[cg];
2563 for (a_gl = gcgs_index[cg_gl]; a_gl < gcgs_index[cg_gl+1]; a_gl++)
2566 ga2la_set(dd->ga2la, a_gl, a, zone1);
2572 gatindex[a] = cg_gl;
2573 ga2la_set(dd->ga2la, cg_gl, a, zone1);
2580 static int check_bLocalCG(gmx_domdec_t *dd, int ncg_sys, const char *bLocalCG,
2583 int ncg, i, ngl, nerr;
2586 if (bLocalCG == NULL)
2590 for (i = 0; i < dd->ncg_tot; i++)
2592 if (!bLocalCG[dd->index_gl[i]])
2595 "DD rank %d, %s: cg %d, global cg %d is not marked in bLocalCG (ncg_home %d)\n", dd->rank, where, i+1, dd->index_gl[i]+1, dd->ncg_home);
2600 for (i = 0; i < ncg_sys; i++)
2607 if (ngl != dd->ncg_tot)
2609 fprintf(stderr, "DD rank %d, %s: In bLocalCG %d cgs are marked as local, whereas there are %d\n", dd->rank, where, ngl, dd->ncg_tot);
2616 static void check_index_consistency(gmx_domdec_t *dd,
2617 int natoms_sys, int ncg_sys,
2620 int nerr, ngl, i, a, cell;
2625 if (dd->comm->DD_debug > 1)
2627 snew(have, natoms_sys);
2628 for (a = 0; a < dd->nat_tot; a++)
2630 if (have[dd->gatindex[a]] > 0)
2632 fprintf(stderr, "DD rank %d: global atom %d occurs twice: index %d and %d\n", dd->rank, dd->gatindex[a]+1, have[dd->gatindex[a]], a+1);
2636 have[dd->gatindex[a]] = a + 1;
2642 snew(have, dd->nat_tot);
2645 for (i = 0; i < natoms_sys; i++)
2647 if (ga2la_get(dd->ga2la, i, &a, &cell))
2649 if (a >= dd->nat_tot)
2651 fprintf(stderr, "DD rank %d: global atom %d marked as local atom %d, which is larger than nat_tot (%d)\n", dd->rank, i+1, a+1, dd->nat_tot);
2657 if (dd->gatindex[a] != i)
2659 fprintf(stderr, "DD rank %d: global atom %d marked as local atom %d, which has global atom index %d\n", dd->rank, i+1, a+1, dd->gatindex[a]+1);
2666 if (ngl != dd->nat_tot)
2669 "DD rank %d, %s: %d global atom indices, %d local atoms\n",
2670 dd->rank, where, ngl, dd->nat_tot);
2672 for (a = 0; a < dd->nat_tot; a++)
2677 "DD rank %d, %s: local atom %d, global %d has no global index\n",
2678 dd->rank, where, a+1, dd->gatindex[a]+1);
2683 nerr += check_bLocalCG(dd, ncg_sys, dd->comm->bLocalCG, where);
2687 gmx_fatal(FARGS, "DD rank %d, %s: %d atom/cg index inconsistencies",
2688 dd->rank, where, nerr);
2692 static void clear_dd_indices(gmx_domdec_t *dd, int cg_start, int a_start)
2699 /* Clear the whole list without searching */
2700 ga2la_clear(dd->ga2la);
2704 for (i = a_start; i < dd->nat_tot; i++)
2706 ga2la_del(dd->ga2la, dd->gatindex[i]);
2710 bLocalCG = dd->comm->bLocalCG;
2713 for (i = cg_start; i < dd->ncg_tot; i++)
2715 bLocalCG[dd->index_gl[i]] = FALSE;
2719 dd_clear_local_vsite_indices(dd);
2721 if (dd->constraints)
2723 dd_clear_local_constraint_indices(dd);
2727 /* This function should be used for moving the domain boudaries during DLB,
2728 * for obtaining the minimum cell size. It checks the initially set limit
2729 * comm->cellsize_min, for bonded and initial non-bonded cut-offs,
2730 * and, possibly, a longer cut-off limit set for PME load balancing.
2732 static real cellsize_min_dlb(gmx_domdec_comm_t *comm, int dim_ind, int dim)
2736 cellsize_min = comm->cellsize_min[dim];
2738 if (!comm->bVacDLBNoLimit)
2740 /* The cut-off might have changed, e.g. by PME load balacning,
2741 * from the value used to set comm->cellsize_min, so check it.
2743 cellsize_min = max(cellsize_min, comm->cutoff/comm->cd[dim_ind].np_dlb);
2745 if (comm->bPMELoadBalDLBLimits)
2747 /* Check for the cut-off limit set by the PME load balancing */
2748 cellsize_min = max(cellsize_min, comm->PMELoadBal_max_cutoff/comm->cd[dim_ind].np_dlb);
2752 return cellsize_min;
2755 static real grid_jump_limit(gmx_domdec_comm_t *comm, real cutoff,
2758 real grid_jump_limit;
2760 /* The distance between the boundaries of cells at distance
2761 * x+-1,y+-1 or y+-1,z+-1 is limited by the cut-off restrictions
2762 * and by the fact that cells should not be shifted by more than
2763 * half their size, such that cg's only shift by one cell
2764 * at redecomposition.
2766 grid_jump_limit = comm->cellsize_limit;
2767 if (!comm->bVacDLBNoLimit)
2769 if (comm->bPMELoadBalDLBLimits)
2771 cutoff = max(cutoff, comm->PMELoadBal_max_cutoff);
2773 grid_jump_limit = max(grid_jump_limit,
2774 cutoff/comm->cd[dim_ind].np);
2777 return grid_jump_limit;
2780 static gmx_bool check_grid_jump(gmx_int64_t step,
2786 gmx_domdec_comm_t *comm;
2795 for (d = 1; d < dd->ndim; d++)
2798 limit = grid_jump_limit(comm, cutoff, d);
2799 bfac = ddbox->box_size[dim];
2800 if (ddbox->tric_dir[dim])
2802 bfac *= ddbox->skew_fac[dim];
2804 if ((comm->cell_f1[d] - comm->cell_f_max0[d])*bfac < limit ||
2805 (comm->cell_f0[d] - comm->cell_f_min1[d])*bfac > -limit)
2813 /* This error should never be triggered under normal
2814 * circumstances, but you never know ...
2816 gmx_fatal(FARGS, "Step %s: The domain decomposition grid has shifted too much in the %c-direction around cell %d %d %d. This should not have happened. Running with fewer ranks might avoid this issue.",
2817 gmx_step_str(step, buf),
2818 dim2char(dim), dd->ci[XX], dd->ci[YY], dd->ci[ZZ]);
2826 static int dd_load_count(gmx_domdec_comm_t *comm)
2828 return (comm->eFlop ? comm->flop_n : comm->cycl_n[ddCyclF]);
2831 static float dd_force_load(gmx_domdec_comm_t *comm)
2838 if (comm->eFlop > 1)
2840 load *= 1.0 + (comm->eFlop - 1)*(0.1*rand()/RAND_MAX - 0.05);
2845 load = comm->cycl[ddCyclF];
2846 if (comm->cycl_n[ddCyclF] > 1)
2848 /* Subtract the maximum of the last n cycle counts
2849 * to get rid of possible high counts due to other sources,
2850 * for instance system activity, that would otherwise
2851 * affect the dynamic load balancing.
2853 load -= comm->cycl_max[ddCyclF];
2857 if (comm->cycl_n[ddCyclWaitGPU] && comm->nrank_gpu_shared > 1)
2859 float gpu_wait, gpu_wait_sum;
2861 gpu_wait = comm->cycl[ddCyclWaitGPU];
2862 if (comm->cycl_n[ddCyclF] > 1)
2864 /* We should remove the WaitGPU time of the same MD step
2865 * as the one with the maximum F time, since the F time
2866 * and the wait time are not independent.
2867 * Furthermore, the step for the max F time should be chosen
2868 * the same on all ranks that share the same GPU.
2869 * But to keep the code simple, we remove the average instead.
2870 * The main reason for artificially long times at some steps
2871 * is spurious CPU activity or MPI time, so we don't expect
2872 * that changes in the GPU wait time matter a lot here.
2874 gpu_wait *= (comm->cycl_n[ddCyclF] - 1)/(float)comm->cycl_n[ddCyclF];
2876 /* Sum the wait times over the ranks that share the same GPU */
2877 MPI_Allreduce(&gpu_wait, &gpu_wait_sum, 1, MPI_FLOAT, MPI_SUM,
2878 comm->mpi_comm_gpu_shared);
2879 /* Replace the wait time by the average over the ranks */
2880 load += -gpu_wait + gpu_wait_sum/comm->nrank_gpu_shared;
2888 static void set_slb_pme_dim_f(gmx_domdec_t *dd, int dim, real **dim_f)
2890 gmx_domdec_comm_t *comm;
2895 snew(*dim_f, dd->nc[dim]+1);
2897 for (i = 1; i < dd->nc[dim]; i++)
2899 if (comm->slb_frac[dim])
2901 (*dim_f)[i] = (*dim_f)[i-1] + comm->slb_frac[dim][i-1];
2905 (*dim_f)[i] = (real)i/(real)dd->nc[dim];
2908 (*dim_f)[dd->nc[dim]] = 1;
2911 static void init_ddpme(gmx_domdec_t *dd, gmx_ddpme_t *ddpme, int dimind)
2913 int pmeindex, slab, nso, i;
2916 if (dimind == 0 && dd->dim[0] == YY && dd->comm->npmenodes_x == 1)
2922 ddpme->dim = dimind;
2924 ddpme->dim_match = (ddpme->dim == dd->dim[dimind]);
2926 ddpme->nslab = (ddpme->dim == 0 ?
2927 dd->comm->npmenodes_x :
2928 dd->comm->npmenodes_y);
2930 if (ddpme->nslab <= 1)
2935 nso = dd->comm->npmenodes/ddpme->nslab;
2936 /* Determine for each PME slab the PP location range for dimension dim */
2937 snew(ddpme->pp_min, ddpme->nslab);
2938 snew(ddpme->pp_max, ddpme->nslab);
2939 for (slab = 0; slab < ddpme->nslab; slab++)
2941 ddpme->pp_min[slab] = dd->nc[dd->dim[dimind]] - 1;
2942 ddpme->pp_max[slab] = 0;
2944 for (i = 0; i < dd->nnodes; i++)
2946 ddindex2xyz(dd->nc, i, xyz);
2947 /* For y only use our y/z slab.
2948 * This assumes that the PME x grid size matches the DD grid size.
2950 if (dimind == 0 || xyz[XX] == dd->ci[XX])
2952 pmeindex = ddindex2pmeindex(dd, i);
2955 slab = pmeindex/nso;
2959 slab = pmeindex % ddpme->nslab;
2961 ddpme->pp_min[slab] = min(ddpme->pp_min[slab], xyz[dimind]);
2962 ddpme->pp_max[slab] = max(ddpme->pp_max[slab], xyz[dimind]);
2966 set_slb_pme_dim_f(dd, ddpme->dim, &ddpme->slb_dim_f);
2969 int dd_pme_maxshift_x(gmx_domdec_t *dd)
2971 if (dd->comm->ddpme[0].dim == XX)
2973 return dd->comm->ddpme[0].maxshift;
2981 int dd_pme_maxshift_y(gmx_domdec_t *dd)
2983 if (dd->comm->ddpme[0].dim == YY)
2985 return dd->comm->ddpme[0].maxshift;
2987 else if (dd->comm->npmedecompdim >= 2 && dd->comm->ddpme[1].dim == YY)
2989 return dd->comm->ddpme[1].maxshift;
2997 static void set_pme_maxshift(gmx_domdec_t *dd, gmx_ddpme_t *ddpme,
2998 gmx_bool bUniform, gmx_ddbox_t *ddbox, real *cell_f)
3000 gmx_domdec_comm_t *comm;
3003 real range, pme_boundary;
3007 nc = dd->nc[ddpme->dim];
3010 if (!ddpme->dim_match)
3012 /* PP decomposition is not along dim: the worst situation */
3015 else if (ns <= 3 || (bUniform && ns == nc))
3017 /* The optimal situation */
3022 /* We need to check for all pme nodes which nodes they
3023 * could possibly need to communicate with.
3025 xmin = ddpme->pp_min;
3026 xmax = ddpme->pp_max;
3027 /* Allow for atoms to be maximally 2/3 times the cut-off
3028 * out of their DD cell. This is a reasonable balance between
3029 * between performance and support for most charge-group/cut-off
3032 range = 2.0/3.0*comm->cutoff/ddbox->box_size[ddpme->dim];
3033 /* Avoid extra communication when we are exactly at a boundary */
3037 for (s = 0; s < ns; s++)
3039 /* PME slab s spreads atoms between box frac. s/ns and (s+1)/ns */
3040 pme_boundary = (real)s/ns;
3043 cell_f[xmax[s-(sh+1) ]+1] + range > pme_boundary) ||
3045 cell_f[xmax[s-(sh+1)+ns]+1] - 1 + range > pme_boundary)))
3049 pme_boundary = (real)(s+1)/ns;
3052 cell_f[xmin[s+(sh+1) ] ] - range < pme_boundary) ||
3054 cell_f[xmin[s+(sh+1)-ns] ] + 1 - range < pme_boundary)))
3061 ddpme->maxshift = sh;
3065 fprintf(debug, "PME slab communication range for dim %d is %d\n",
3066 ddpme->dim, ddpme->maxshift);
3070 static void check_box_size(gmx_domdec_t *dd, gmx_ddbox_t *ddbox)
3074 for (d = 0; d < dd->ndim; d++)
3077 if (dim < ddbox->nboundeddim &&
3078 ddbox->box_size[dim]*ddbox->skew_fac[dim] <
3079 dd->nc[dim]*dd->comm->cellsize_limit*DD_CELL_MARGIN)
3081 gmx_fatal(FARGS, "The %c-size of the box (%f) times the triclinic skew factor (%f) is smaller than the number of DD cells (%d) times the smallest allowed cell size (%f)\n",
3082 dim2char(dim), ddbox->box_size[dim], ddbox->skew_fac[dim],
3083 dd->nc[dim], dd->comm->cellsize_limit);
3089 setcellsizeslbLOCAL, setcellsizeslbMASTER, setcellsizeslbPULSE_ONLY
3092 /* Set the domain boundaries. Use for static (or no) load balancing,
3093 * and also for the starting state for dynamic load balancing.
3094 * setmode determine if and where the boundaries are stored, use enum above.
3095 * Returns the number communication pulses in npulse.
3097 static void set_dd_cell_sizes_slb(gmx_domdec_t *dd, gmx_ddbox_t *ddbox,
3098 int setmode, ivec npulse)
3100 gmx_domdec_comm_t *comm;
3103 real *cell_x, cell_dx, cellsize;
3107 for (d = 0; d < DIM; d++)
3109 cellsize_min[d] = ddbox->box_size[d]*ddbox->skew_fac[d];
3111 if (dd->nc[d] == 1 || comm->slb_frac[d] == NULL)
3114 cell_dx = ddbox->box_size[d]/dd->nc[d];
3117 case setcellsizeslbMASTER:
3118 for (j = 0; j < dd->nc[d]+1; j++)
3120 dd->ma->cell_x[d][j] = ddbox->box0[d] + j*cell_dx;
3123 case setcellsizeslbLOCAL:
3124 comm->cell_x0[d] = ddbox->box0[d] + (dd->ci[d] )*cell_dx;
3125 comm->cell_x1[d] = ddbox->box0[d] + (dd->ci[d]+1)*cell_dx;
3130 cellsize = cell_dx*ddbox->skew_fac[d];
3131 while (cellsize*npulse[d] < comm->cutoff)
3135 cellsize_min[d] = cellsize;
3139 /* Statically load balanced grid */
3140 /* Also when we are not doing a master distribution we determine
3141 * all cell borders in a loop to obtain identical values
3142 * to the master distribution case and to determine npulse.
3144 if (setmode == setcellsizeslbMASTER)
3146 cell_x = dd->ma->cell_x[d];
3150 snew(cell_x, dd->nc[d]+1);
3152 cell_x[0] = ddbox->box0[d];
3153 for (j = 0; j < dd->nc[d]; j++)
3155 cell_dx = ddbox->box_size[d]*comm->slb_frac[d][j];
3156 cell_x[j+1] = cell_x[j] + cell_dx;
3157 cellsize = cell_dx*ddbox->skew_fac[d];
3158 while (cellsize*npulse[d] < comm->cutoff &&
3159 npulse[d] < dd->nc[d]-1)
3163 cellsize_min[d] = min(cellsize_min[d], cellsize);
3165 if (setmode == setcellsizeslbLOCAL)
3167 comm->cell_x0[d] = cell_x[dd->ci[d]];
3168 comm->cell_x1[d] = cell_x[dd->ci[d]+1];
3170 if (setmode != setcellsizeslbMASTER)
3175 /* The following limitation is to avoid that a cell would receive
3176 * some of its own home charge groups back over the periodic boundary.
3177 * Double charge groups cause trouble with the global indices.
3179 if (d < ddbox->npbcdim &&
3180 dd->nc[d] > 1 && npulse[d] >= dd->nc[d])
3182 char error_string[STRLEN];
3184 sprintf(error_string,
3185 "The box size in direction %c (%f) times the triclinic skew factor (%f) is too small for a cut-off of %f with %d domain decomposition cells, use 1 or more than %d %s or increase the box size in this direction",
3186 dim2char(d), ddbox->box_size[d], ddbox->skew_fac[d],
3188 dd->nc[d], dd->nc[d],
3189 dd->nnodes > dd->nc[d] ? "cells" : "ranks");
3191 if (setmode == setcellsizeslbLOCAL)
3193 gmx_fatal_collective(FARGS, NULL, dd, error_string);
3197 gmx_fatal(FARGS, error_string);
3202 if (!comm->bDynLoadBal)
3204 copy_rvec(cellsize_min, comm->cellsize_min);
3207 for (d = 0; d < comm->npmedecompdim; d++)
3209 set_pme_maxshift(dd, &comm->ddpme[d],
3210 comm->slb_frac[dd->dim[d]] == NULL, ddbox,
3211 comm->ddpme[d].slb_dim_f);
3216 static void dd_cell_sizes_dlb_root_enforce_limits(gmx_domdec_t *dd,
3217 int d, int dim, gmx_domdec_root_t *root,
3219 gmx_bool bUniform, gmx_int64_t step, real cellsize_limit_f, int range[])
3221 gmx_domdec_comm_t *comm;
3222 int ncd, i, j, nmin, nmin_old;
3223 gmx_bool bLimLo, bLimHi;
3225 real fac, halfway, cellsize_limit_f_i, region_size;
3226 gmx_bool bPBC, bLastHi = FALSE;
3227 int nrange[] = {range[0], range[1]};
3229 region_size = root->cell_f[range[1]]-root->cell_f[range[0]];
3235 bPBC = (dim < ddbox->npbcdim);
3237 cell_size = root->buf_ncd;
3241 fprintf(debug, "enforce_limits: %d %d\n", range[0], range[1]);
3244 /* First we need to check if the scaling does not make cells
3245 * smaller than the smallest allowed size.
3246 * We need to do this iteratively, since if a cell is too small,
3247 * it needs to be enlarged, which makes all the other cells smaller,
3248 * which could in turn make another cell smaller than allowed.
3250 for (i = range[0]; i < range[1]; i++)
3252 root->bCellMin[i] = FALSE;
3258 /* We need the total for normalization */
3260 for (i = range[0]; i < range[1]; i++)
3262 if (root->bCellMin[i] == FALSE)
3264 fac += cell_size[i];
3267 fac = ( region_size - nmin*cellsize_limit_f)/fac; /* substracting cells already set to cellsize_limit_f */
3268 /* Determine the cell boundaries */
3269 for (i = range[0]; i < range[1]; i++)
3271 if (root->bCellMin[i] == FALSE)
3273 cell_size[i] *= fac;
3274 if (!bPBC && (i == 0 || i == dd->nc[dim] -1))
3276 cellsize_limit_f_i = 0;
3280 cellsize_limit_f_i = cellsize_limit_f;
3282 if (cell_size[i] < cellsize_limit_f_i)
3284 root->bCellMin[i] = TRUE;
3285 cell_size[i] = cellsize_limit_f_i;
3289 root->cell_f[i+1] = root->cell_f[i] + cell_size[i];
3292 while (nmin > nmin_old);
3295 cell_size[i] = root->cell_f[i+1] - root->cell_f[i];
3296 /* For this check we should not use DD_CELL_MARGIN,
3297 * but a slightly smaller factor,
3298 * since rounding could get use below the limit.
3300 if (bPBC && cell_size[i] < cellsize_limit_f*DD_CELL_MARGIN2/DD_CELL_MARGIN)
3303 gmx_fatal(FARGS, "Step %s: the dynamic load balancing could not balance dimension %c: box size %f, triclinic skew factor %f, #cells %d, minimum cell size %f\n",
3304 gmx_step_str(step, buf),
3305 dim2char(dim), ddbox->box_size[dim], ddbox->skew_fac[dim],
3306 ncd, comm->cellsize_min[dim]);
3309 root->bLimited = (nmin > 0) || (range[0] > 0) || (range[1] < ncd);
3313 /* Check if the boundary did not displace more than halfway
3314 * each of the cells it bounds, as this could cause problems,
3315 * especially when the differences between cell sizes are large.
3316 * If changes are applied, they will not make cells smaller
3317 * than the cut-off, as we check all the boundaries which
3318 * might be affected by a change and if the old state was ok,
3319 * the cells will at most be shrunk back to their old size.
3321 for (i = range[0]+1; i < range[1]; i++)
3323 halfway = 0.5*(root->old_cell_f[i] + root->old_cell_f[i-1]);
3324 if (root->cell_f[i] < halfway)
3326 root->cell_f[i] = halfway;
3327 /* Check if the change also causes shifts of the next boundaries */
3328 for (j = i+1; j < range[1]; j++)
3330 if (root->cell_f[j] < root->cell_f[j-1] + cellsize_limit_f)
3332 root->cell_f[j] = root->cell_f[j-1] + cellsize_limit_f;
3336 halfway = 0.5*(root->old_cell_f[i] + root->old_cell_f[i+1]);
3337 if (root->cell_f[i] > halfway)
3339 root->cell_f[i] = halfway;
3340 /* Check if the change also causes shifts of the next boundaries */
3341 for (j = i-1; j >= range[0]+1; j--)
3343 if (root->cell_f[j] > root->cell_f[j+1] - cellsize_limit_f)
3345 root->cell_f[j] = root->cell_f[j+1] - cellsize_limit_f;
3352 /* nrange is defined as [lower, upper) range for new call to enforce_limits */
3353 /* find highest violation of LimLo (a) and the following violation of LimHi (thus the lowest following) (b)
3354 * then call enforce_limits for (oldb,a), (a,b). In the next step: (b,nexta). oldb and nexta can be the boundaries.
3355 * for a and b nrange is used */
3358 /* Take care of the staggering of the cell boundaries */
3361 for (i = range[0]; i < range[1]; i++)
3363 root->cell_f_max0[i] = root->cell_f[i];
3364 root->cell_f_min1[i] = root->cell_f[i+1];
3369 for (i = range[0]+1; i < range[1]; i++)
3371 bLimLo = (root->cell_f[i] < root->bound_min[i]);
3372 bLimHi = (root->cell_f[i] > root->bound_max[i]);
3373 if (bLimLo && bLimHi)
3375 /* Both limits violated, try the best we can */
3376 /* For this case we split the original range (range) in two parts and care about the other limitiations in the next iteration. */
3377 root->cell_f[i] = 0.5*(root->bound_min[i] + root->bound_max[i]);
3378 nrange[0] = range[0];
3380 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, nrange);
3383 nrange[1] = range[1];
3384 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, nrange);
3390 /* root->cell_f[i] = root->bound_min[i]; */
3391 nrange[1] = i; /* only store violation location. There could be a LimLo violation following with an higher index */
3394 else if (bLimHi && !bLastHi)
3397 if (nrange[1] < range[1]) /* found a LimLo before */
3399 root->cell_f[nrange[1]] = root->bound_min[nrange[1]];
3400 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, nrange);
3401 nrange[0] = nrange[1];
3403 root->cell_f[i] = root->bound_max[i];
3405 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, nrange);
3407 nrange[1] = range[1];
3410 if (nrange[1] < range[1]) /* found last a LimLo */
3412 root->cell_f[nrange[1]] = root->bound_min[nrange[1]];
3413 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, nrange);
3414 nrange[0] = nrange[1];
3415 nrange[1] = range[1];
3416 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, nrange);
3418 else if (nrange[0] > range[0]) /* found at least one LimHi */
3420 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, nrange);
3427 static void set_dd_cell_sizes_dlb_root(gmx_domdec_t *dd,
3428 int d, int dim, gmx_domdec_root_t *root,
3429 gmx_ddbox_t *ddbox, gmx_bool bDynamicBox,
3430 gmx_bool bUniform, gmx_int64_t step)
3432 gmx_domdec_comm_t *comm;
3433 int ncd, d1, i, j, pos;
3435 real load_aver, load_i, imbalance, change, change_max, sc;
3436 real cellsize_limit_f, dist_min_f, dist_min_f_hard, space;
3440 int range[] = { 0, 0 };
3444 /* Convert the maximum change from the input percentage to a fraction */
3445 change_limit = comm->dlb_scale_lim*0.01;
3449 bPBC = (dim < ddbox->npbcdim);
3451 cell_size = root->buf_ncd;
3453 /* Store the original boundaries */
3454 for (i = 0; i < ncd+1; i++)
3456 root->old_cell_f[i] = root->cell_f[i];
3460 for (i = 0; i < ncd; i++)
3462 cell_size[i] = 1.0/ncd;
3465 else if (dd_load_count(comm))
3467 load_aver = comm->load[d].sum_m/ncd;
3469 for (i = 0; i < ncd; i++)
3471 /* Determine the relative imbalance of cell i */
3472 load_i = comm->load[d].load[i*comm->load[d].nload+2];
3473 imbalance = (load_i - load_aver)/(load_aver > 0 ? load_aver : 1);
3474 /* Determine the change of the cell size using underrelaxation */
3475 change = -relax*imbalance;
3476 change_max = max(change_max, max(change, -change));
3478 /* Limit the amount of scaling.
3479 * We need to use the same rescaling for all cells in one row,
3480 * otherwise the load balancing might not converge.
3483 if (change_max > change_limit)
3485 sc *= change_limit/change_max;
3487 for (i = 0; i < ncd; i++)
3489 /* Determine the relative imbalance of cell i */
3490 load_i = comm->load[d].load[i*comm->load[d].nload+2];
3491 imbalance = (load_i - load_aver)/(load_aver > 0 ? load_aver : 1);
3492 /* Determine the change of the cell size using underrelaxation */
3493 change = -sc*imbalance;
3494 cell_size[i] = (root->cell_f[i+1]-root->cell_f[i])*(1 + change);
3498 cellsize_limit_f = cellsize_min_dlb(comm, d, dim)/ddbox->box_size[dim];
3499 cellsize_limit_f *= DD_CELL_MARGIN;
3500 dist_min_f_hard = grid_jump_limit(comm, comm->cutoff, d)/ddbox->box_size[dim];
3501 dist_min_f = dist_min_f_hard * DD_CELL_MARGIN;
3502 if (ddbox->tric_dir[dim])
3504 cellsize_limit_f /= ddbox->skew_fac[dim];
3505 dist_min_f /= ddbox->skew_fac[dim];
3507 if (bDynamicBox && d > 0)
3509 dist_min_f *= DD_PRES_SCALE_MARGIN;
3511 if (d > 0 && !bUniform)
3513 /* Make sure that the grid is not shifted too much */
3514 for (i = 1; i < ncd; i++)
3516 if (root->cell_f_min1[i] - root->cell_f_max0[i-1] < 2 * dist_min_f_hard)
3518 gmx_incons("Inconsistent DD boundary staggering limits!");
3520 root->bound_min[i] = root->cell_f_max0[i-1] + dist_min_f;
3521 space = root->cell_f[i] - (root->cell_f_max0[i-1] + dist_min_f);
3524 root->bound_min[i] += 0.5*space;
3526 root->bound_max[i] = root->cell_f_min1[i] - dist_min_f;
3527 space = root->cell_f[i] - (root->cell_f_min1[i] - dist_min_f);
3530 root->bound_max[i] += 0.5*space;
3535 "dim %d boundary %d %.3f < %.3f < %.3f < %.3f < %.3f\n",
3537 root->cell_f_max0[i-1] + dist_min_f,
3538 root->bound_min[i], root->cell_f[i], root->bound_max[i],
3539 root->cell_f_min1[i] - dist_min_f);
3544 root->cell_f[0] = 0;
3545 root->cell_f[ncd] = 1;
3546 dd_cell_sizes_dlb_root_enforce_limits(dd, d, dim, root, ddbox, bUniform, step, cellsize_limit_f, range);
3549 /* After the checks above, the cells should obey the cut-off
3550 * restrictions, but it does not hurt to check.
3552 for (i = 0; i < ncd; i++)
3556 fprintf(debug, "Relative bounds dim %d cell %d: %f %f\n",
3557 dim, i, root->cell_f[i], root->cell_f[i+1]);
3560 if ((bPBC || (i != 0 && i != dd->nc[dim]-1)) &&
3561 root->cell_f[i+1] - root->cell_f[i] <
3562 cellsize_limit_f/DD_CELL_MARGIN)
3566 "\nWARNING step %s: direction %c, cell %d too small: %f\n",
3567 gmx_step_str(step, buf), dim2char(dim), i,
3568 (root->cell_f[i+1] - root->cell_f[i])
3569 *ddbox->box_size[dim]*ddbox->skew_fac[dim]);
3574 /* Store the cell boundaries of the lower dimensions at the end */
3575 for (d1 = 0; d1 < d; d1++)
3577 root->cell_f[pos++] = comm->cell_f0[d1];
3578 root->cell_f[pos++] = comm->cell_f1[d1];
3581 if (d < comm->npmedecompdim)
3583 /* The master determines the maximum shift for
3584 * the coordinate communication between separate PME nodes.
3586 set_pme_maxshift(dd, &comm->ddpme[d], bUniform, ddbox, root->cell_f);
3588 root->cell_f[pos++] = comm->ddpme[0].maxshift;
3591 root->cell_f[pos++] = comm->ddpme[1].maxshift;
3595 static void relative_to_absolute_cell_bounds(gmx_domdec_t *dd,
3596 gmx_ddbox_t *ddbox, int dimind)
3598 gmx_domdec_comm_t *comm;
3603 /* Set the cell dimensions */
3604 dim = dd->dim[dimind];
3605 comm->cell_x0[dim] = comm->cell_f0[dimind]*ddbox->box_size[dim];
3606 comm->cell_x1[dim] = comm->cell_f1[dimind]*ddbox->box_size[dim];
3607 if (dim >= ddbox->nboundeddim)
3609 comm->cell_x0[dim] += ddbox->box0[dim];
3610 comm->cell_x1[dim] += ddbox->box0[dim];
3614 static void distribute_dd_cell_sizes_dlb(gmx_domdec_t *dd,
3615 int d, int dim, real *cell_f_row,
3618 gmx_domdec_comm_t *comm;
3624 /* Each node would only need to know two fractions,
3625 * but it is probably cheaper to broadcast the whole array.
3627 MPI_Bcast(cell_f_row, DD_CELL_F_SIZE(dd, d)*sizeof(real), MPI_BYTE,
3628 0, comm->mpi_comm_load[d]);
3630 /* Copy the fractions for this dimension from the buffer */
3631 comm->cell_f0[d] = cell_f_row[dd->ci[dim] ];
3632 comm->cell_f1[d] = cell_f_row[dd->ci[dim]+1];
3633 /* The whole array was communicated, so set the buffer position */
3634 pos = dd->nc[dim] + 1;
3635 for (d1 = 0; d1 <= d; d1++)
3639 /* Copy the cell fractions of the lower dimensions */
3640 comm->cell_f0[d1] = cell_f_row[pos++];
3641 comm->cell_f1[d1] = cell_f_row[pos++];
3643 relative_to_absolute_cell_bounds(dd, ddbox, d1);
3645 /* Convert the communicated shift from float to int */
3646 comm->ddpme[0].maxshift = (int)(cell_f_row[pos++] + 0.5);
3649 comm->ddpme[1].maxshift = (int)(cell_f_row[pos++] + 0.5);
3653 static void set_dd_cell_sizes_dlb_change(gmx_domdec_t *dd,
3654 gmx_ddbox_t *ddbox, gmx_bool bDynamicBox,
3655 gmx_bool bUniform, gmx_int64_t step)
3657 gmx_domdec_comm_t *comm;
3659 gmx_bool bRowMember, bRowRoot;
3664 for (d = 0; d < dd->ndim; d++)
3669 for (d1 = d; d1 < dd->ndim; d1++)
3671 if (dd->ci[dd->dim[d1]] > 0)
3684 set_dd_cell_sizes_dlb_root(dd, d, dim, comm->root[d],
3685 ddbox, bDynamicBox, bUniform, step);
3686 cell_f_row = comm->root[d]->cell_f;
3690 cell_f_row = comm->cell_f_row;
3692 distribute_dd_cell_sizes_dlb(dd, d, dim, cell_f_row, ddbox);
3697 static void set_dd_cell_sizes_dlb_nochange(gmx_domdec_t *dd, gmx_ddbox_t *ddbox)
3701 /* This function assumes the box is static and should therefore
3702 * not be called when the box has changed since the last
3703 * call to dd_partition_system.
3705 for (d = 0; d < dd->ndim; d++)
3707 relative_to_absolute_cell_bounds(dd, ddbox, d);
3713 static void set_dd_cell_sizes_dlb(gmx_domdec_t *dd,
3714 gmx_ddbox_t *ddbox, gmx_bool bDynamicBox,
3715 gmx_bool bUniform, gmx_bool bDoDLB, gmx_int64_t step,
3716 gmx_wallcycle_t wcycle)
3718 gmx_domdec_comm_t *comm;
3725 wallcycle_start(wcycle, ewcDDCOMMBOUND);
3726 set_dd_cell_sizes_dlb_change(dd, ddbox, bDynamicBox, bUniform, step);
3727 wallcycle_stop(wcycle, ewcDDCOMMBOUND);
3729 else if (bDynamicBox)
3731 set_dd_cell_sizes_dlb_nochange(dd, ddbox);
3734 /* Set the dimensions for which no DD is used */
3735 for (dim = 0; dim < DIM; dim++)
3737 if (dd->nc[dim] == 1)
3739 comm->cell_x0[dim] = 0;
3740 comm->cell_x1[dim] = ddbox->box_size[dim];
3741 if (dim >= ddbox->nboundeddim)
3743 comm->cell_x0[dim] += ddbox->box0[dim];
3744 comm->cell_x1[dim] += ddbox->box0[dim];
3750 static void realloc_comm_ind(gmx_domdec_t *dd, ivec npulse)
3753 gmx_domdec_comm_dim_t *cd;
3755 for (d = 0; d < dd->ndim; d++)
3757 cd = &dd->comm->cd[d];
3758 np = npulse[dd->dim[d]];
3759 if (np > cd->np_nalloc)
3763 fprintf(debug, "(Re)allocing cd for %c to %d pulses\n",
3764 dim2char(dd->dim[d]), np);
3766 if (DDMASTER(dd) && cd->np_nalloc > 0)
3768 fprintf(stderr, "\nIncreasing the number of cell to communicate in dimension %c to %d for the first time\n", dim2char(dd->dim[d]), np);
3770 srenew(cd->ind, np);
3771 for (i = cd->np_nalloc; i < np; i++)
3773 cd->ind[i].index = NULL;
3774 cd->ind[i].nalloc = 0;
3783 static void set_dd_cell_sizes(gmx_domdec_t *dd,
3784 gmx_ddbox_t *ddbox, gmx_bool bDynamicBox,
3785 gmx_bool bUniform, gmx_bool bDoDLB, gmx_int64_t step,
3786 gmx_wallcycle_t wcycle)
3788 gmx_domdec_comm_t *comm;
3794 /* Copy the old cell boundaries for the cg displacement check */
3795 copy_rvec(comm->cell_x0, comm->old_cell_x0);
3796 copy_rvec(comm->cell_x1, comm->old_cell_x1);
3798 if (comm->bDynLoadBal)
3802 check_box_size(dd, ddbox);
3804 set_dd_cell_sizes_dlb(dd, ddbox, bDynamicBox, bUniform, bDoDLB, step, wcycle);
3808 set_dd_cell_sizes_slb(dd, ddbox, setcellsizeslbLOCAL, npulse);
3809 realloc_comm_ind(dd, npulse);
3814 for (d = 0; d < DIM; d++)
3816 fprintf(debug, "cell_x[%d] %f - %f skew_fac %f\n",
3817 d, comm->cell_x0[d], comm->cell_x1[d], ddbox->skew_fac[d]);
3822 static void comm_dd_ns_cell_sizes(gmx_domdec_t *dd,
3824 rvec cell_ns_x0, rvec cell_ns_x1,
3827 gmx_domdec_comm_t *comm;
3832 for (dim_ind = 0; dim_ind < dd->ndim; dim_ind++)
3834 dim = dd->dim[dim_ind];
3836 /* Without PBC we don't have restrictions on the outer cells */
3837 if (!(dim >= ddbox->npbcdim &&
3838 (dd->ci[dim] == 0 || dd->ci[dim] == dd->nc[dim] - 1)) &&
3839 comm->bDynLoadBal &&
3840 (comm->cell_x1[dim] - comm->cell_x0[dim])*ddbox->skew_fac[dim] <
3841 comm->cellsize_min[dim])
3844 gmx_fatal(FARGS, "Step %s: The %c-size (%f) times the triclinic skew factor (%f) is smaller than the smallest allowed cell size (%f) for domain decomposition grid cell %d %d %d",
3845 gmx_step_str(step, buf), dim2char(dim),
3846 comm->cell_x1[dim] - comm->cell_x0[dim],
3847 ddbox->skew_fac[dim],
3848 dd->comm->cellsize_min[dim],
3849 dd->ci[XX], dd->ci[YY], dd->ci[ZZ]);
3853 if ((dd->bGridJump && dd->ndim > 1) || ddbox->nboundeddim < DIM)
3855 /* Communicate the boundaries and update cell_ns_x0/1 */
3856 dd_move_cellx(dd, ddbox, cell_ns_x0, cell_ns_x1);
3857 if (dd->bGridJump && dd->ndim > 1)
3859 check_grid_jump(step, dd, dd->comm->cutoff, ddbox, TRUE);
3864 static void make_tric_corr_matrix(int npbcdim, matrix box, matrix tcm)
3868 tcm[YY][XX] = -box[YY][XX]/box[YY][YY];
3876 tcm[ZZ][XX] = -(box[ZZ][YY]*tcm[YY][XX] + box[ZZ][XX])/box[ZZ][ZZ];
3877 tcm[ZZ][YY] = -box[ZZ][YY]/box[ZZ][ZZ];
3886 static void check_screw_box(matrix box)
3888 /* Mathematical limitation */
3889 if (box[YY][XX] != 0 || box[ZZ][XX] != 0)
3891 gmx_fatal(FARGS, "With screw pbc the unit cell can not have non-zero off-diagonal x-components");
3894 /* Limitation due to the asymmetry of the eighth shell method */
3895 if (box[ZZ][YY] != 0)
3897 gmx_fatal(FARGS, "pbc=screw with non-zero box_zy is not supported");
3901 static void distribute_cg(FILE *fplog, gmx_int64_t step,
3902 matrix box, ivec tric_dir, t_block *cgs, rvec pos[],
3905 gmx_domdec_master_t *ma;
3906 int **tmp_ind = NULL, *tmp_nalloc = NULL;
3907 int i, icg, j, k, k0, k1, d, npbcdim;
3909 rvec box_size, cg_cm;
3911 real nrcg, inv_ncg, pos_d;
3913 gmx_bool bUnbounded, bScrew;
3917 if (tmp_ind == NULL)
3919 snew(tmp_nalloc, dd->nnodes);
3920 snew(tmp_ind, dd->nnodes);
3921 for (i = 0; i < dd->nnodes; i++)
3923 tmp_nalloc[i] = over_alloc_large(cgs->nr/dd->nnodes+1);
3924 snew(tmp_ind[i], tmp_nalloc[i]);
3928 /* Clear the count */
3929 for (i = 0; i < dd->nnodes; i++)
3935 make_tric_corr_matrix(dd->npbcdim, box, tcm);
3937 cgindex = cgs->index;
3939 /* Compute the center of geometry for all charge groups */
3940 for (icg = 0; icg < cgs->nr; icg++)
3943 k1 = cgindex[icg+1];
3947 copy_rvec(pos[k0], cg_cm);
3954 for (k = k0; (k < k1); k++)
3956 rvec_inc(cg_cm, pos[k]);
3958 for (d = 0; (d < DIM); d++)
3960 cg_cm[d] *= inv_ncg;
3963 /* Put the charge group in the box and determine the cell index */
3964 for (d = DIM-1; d >= 0; d--)
3967 if (d < dd->npbcdim)
3969 bScrew = (dd->bScrewPBC && d == XX);
3970 if (tric_dir[d] && dd->nc[d] > 1)
3972 /* Use triclinic coordintates for this dimension */
3973 for (j = d+1; j < DIM; j++)
3975 pos_d += cg_cm[j]*tcm[j][d];
3978 while (pos_d >= box[d][d])
3981 rvec_dec(cg_cm, box[d]);
3984 cg_cm[YY] = box[YY][YY] - cg_cm[YY];
3985 cg_cm[ZZ] = box[ZZ][ZZ] - cg_cm[ZZ];
3987 for (k = k0; (k < k1); k++)
3989 rvec_dec(pos[k], box[d]);
3992 pos[k][YY] = box[YY][YY] - pos[k][YY];
3993 pos[k][ZZ] = box[ZZ][ZZ] - pos[k][ZZ];
4000 rvec_inc(cg_cm, box[d]);
4003 cg_cm[YY] = box[YY][YY] - cg_cm[YY];
4004 cg_cm[ZZ] = box[ZZ][ZZ] - cg_cm[ZZ];
4006 for (k = k0; (k < k1); k++)
4008 rvec_inc(pos[k], box[d]);
4011 pos[k][YY] = box[YY][YY] - pos[k][YY];
4012 pos[k][ZZ] = box[ZZ][ZZ] - pos[k][ZZ];
4017 /* This could be done more efficiently */
4019 while (ind[d]+1 < dd->nc[d] && pos_d >= ma->cell_x[d][ind[d]+1])
4024 i = dd_index(dd->nc, ind);
4025 if (ma->ncg[i] == tmp_nalloc[i])
4027 tmp_nalloc[i] = over_alloc_large(ma->ncg[i]+1);
4028 srenew(tmp_ind[i], tmp_nalloc[i]);
4030 tmp_ind[i][ma->ncg[i]] = icg;
4032 ma->nat[i] += cgindex[icg+1] - cgindex[icg];
4036 for (i = 0; i < dd->nnodes; i++)
4039 for (k = 0; k < ma->ncg[i]; k++)
4041 ma->cg[k1++] = tmp_ind[i][k];
4044 ma->index[dd->nnodes] = k1;
4046 for (i = 0; i < dd->nnodes; i++)
4056 fprintf(fplog, "Charge group distribution at step %s:",
4057 gmx_step_str(step, buf));
4058 for (i = 0; i < dd->nnodes; i++)
4060 fprintf(fplog, " %d", ma->ncg[i]);
4062 fprintf(fplog, "\n");
4066 static void get_cg_distribution(FILE *fplog, gmx_int64_t step, gmx_domdec_t *dd,
4067 t_block *cgs, matrix box, gmx_ddbox_t *ddbox,
4070 gmx_domdec_master_t *ma = NULL;
4073 int *ibuf, buf2[2] = { 0, 0 };
4074 gmx_bool bMaster = DDMASTER(dd);
4082 check_screw_box(box);
4085 set_dd_cell_sizes_slb(dd, ddbox, setcellsizeslbMASTER, npulse);
4087 distribute_cg(fplog, step, box, ddbox->tric_dir, cgs, pos, dd);
4088 for (i = 0; i < dd->nnodes; i++)
4090 ma->ibuf[2*i] = ma->ncg[i];
4091 ma->ibuf[2*i+1] = ma->nat[i];
4099 dd_scatter(dd, 2*sizeof(int), ibuf, buf2);
4101 dd->ncg_home = buf2[0];
4102 dd->nat_home = buf2[1];
4103 dd->ncg_tot = dd->ncg_home;
4104 dd->nat_tot = dd->nat_home;
4105 if (dd->ncg_home > dd->cg_nalloc || dd->cg_nalloc == 0)
4107 dd->cg_nalloc = over_alloc_dd(dd->ncg_home);
4108 srenew(dd->index_gl, dd->cg_nalloc);
4109 srenew(dd->cgindex, dd->cg_nalloc+1);
4113 for (i = 0; i < dd->nnodes; i++)
4115 ma->ibuf[i] = ma->ncg[i]*sizeof(int);
4116 ma->ibuf[dd->nnodes+i] = ma->index[i]*sizeof(int);
4121 DDMASTER(dd) ? ma->ibuf : NULL,
4122 DDMASTER(dd) ? ma->ibuf+dd->nnodes : NULL,
4123 DDMASTER(dd) ? ma->cg : NULL,
4124 dd->ncg_home*sizeof(int), dd->index_gl);
4126 /* Determine the home charge group sizes */
4128 for (i = 0; i < dd->ncg_home; i++)
4130 cg_gl = dd->index_gl[i];
4132 dd->cgindex[i] + cgs->index[cg_gl+1] - cgs->index[cg_gl];
4137 fprintf(debug, "Home charge groups:\n");
4138 for (i = 0; i < dd->ncg_home; i++)
4140 fprintf(debug, " %d", dd->index_gl[i]);
4143 fprintf(debug, "\n");
4146 fprintf(debug, "\n");
4150 static int compact_and_copy_vec_at(int ncg, int *move,
4153 rvec *src, gmx_domdec_comm_t *comm,
4156 int m, icg, i, i0, i1, nrcg;
4162 for (m = 0; m < DIM*2; m++)
4168 for (icg = 0; icg < ncg; icg++)
4170 i1 = cgindex[icg+1];
4176 /* Compact the home array in place */
4177 for (i = i0; i < i1; i++)
4179 copy_rvec(src[i], src[home_pos++]);
4185 /* Copy to the communication buffer */
4187 pos_vec[m] += 1 + vec*nrcg;
4188 for (i = i0; i < i1; i++)
4190 copy_rvec(src[i], comm->cgcm_state[m][pos_vec[m]++]);
4192 pos_vec[m] += (nvec - vec - 1)*nrcg;
4196 home_pos += i1 - i0;
4204 static int compact_and_copy_vec_cg(int ncg, int *move,
4206 int nvec, rvec *src, gmx_domdec_comm_t *comm,
4209 int m, icg, i0, i1, nrcg;
4215 for (m = 0; m < DIM*2; m++)
4221 for (icg = 0; icg < ncg; icg++)
4223 i1 = cgindex[icg+1];
4229 /* Compact the home array in place */
4230 copy_rvec(src[icg], src[home_pos++]);
4236 /* Copy to the communication buffer */
4237 copy_rvec(src[icg], comm->cgcm_state[m][pos_vec[m]]);
4238 pos_vec[m] += 1 + nrcg*nvec;
4250 static int compact_ind(int ncg, int *move,
4251 int *index_gl, int *cgindex,
4253 gmx_ga2la_t ga2la, char *bLocalCG,
4256 int cg, nat, a0, a1, a, a_gl;
4261 for (cg = 0; cg < ncg; cg++)
4267 /* Compact the home arrays in place.
4268 * Anything that can be done here avoids access to global arrays.
4270 cgindex[home_pos] = nat;
4271 for (a = a0; a < a1; a++)
4274 gatindex[nat] = a_gl;
4275 /* The cell number stays 0, so we don't need to set it */
4276 ga2la_change_la(ga2la, a_gl, nat);
4279 index_gl[home_pos] = index_gl[cg];
4280 cginfo[home_pos] = cginfo[cg];
4281 /* The charge group remains local, so bLocalCG does not change */
4286 /* Clear the global indices */
4287 for (a = a0; a < a1; a++)
4289 ga2la_del(ga2la, gatindex[a]);
4293 bLocalCG[index_gl[cg]] = FALSE;
4297 cgindex[home_pos] = nat;
4302 static void clear_and_mark_ind(int ncg, int *move,
4303 int *index_gl, int *cgindex, int *gatindex,
4304 gmx_ga2la_t ga2la, char *bLocalCG,
4309 for (cg = 0; cg < ncg; cg++)
4315 /* Clear the global indices */
4316 for (a = a0; a < a1; a++)
4318 ga2la_del(ga2la, gatindex[a]);
4322 bLocalCG[index_gl[cg]] = FALSE;
4324 /* Signal that this cg has moved using the ns cell index.
4325 * Here we set it to -1. fill_grid will change it
4326 * from -1 to NSGRID_SIGNAL_MOVED_FAC*grid->ncells.
4328 cell_index[cg] = -1;
4333 static void print_cg_move(FILE *fplog,
4335 gmx_int64_t step, int cg, int dim, int dir,
4336 gmx_bool bHaveLimitdAndCMOld, real limitd,
4337 rvec cm_old, rvec cm_new, real pos_d)
4339 gmx_domdec_comm_t *comm;
4344 fprintf(fplog, "\nStep %s:\n", gmx_step_str(step, buf));
4345 if (bHaveLimitdAndCMOld)
4347 fprintf(fplog, "The charge group starting at atom %d moved more than the distance allowed by the domain decomposition (%f) in direction %c\n",
4348 ddglatnr(dd, dd->cgindex[cg]), limitd, dim2char(dim));
4352 fprintf(fplog, "The charge group starting at atom %d moved than the distance allowed by the domain decomposition in direction %c\n",
4353 ddglatnr(dd, dd->cgindex[cg]), dim2char(dim));
4355 fprintf(fplog, "distance out of cell %f\n",
4356 dir == 1 ? pos_d - comm->cell_x1[dim] : pos_d - comm->cell_x0[dim]);
4357 if (bHaveLimitdAndCMOld)
4359 fprintf(fplog, "Old coordinates: %8.3f %8.3f %8.3f\n",
4360 cm_old[XX], cm_old[YY], cm_old[ZZ]);
4362 fprintf(fplog, "New coordinates: %8.3f %8.3f %8.3f\n",
4363 cm_new[XX], cm_new[YY], cm_new[ZZ]);
4364 fprintf(fplog, "Old cell boundaries in direction %c: %8.3f %8.3f\n",
4366 comm->old_cell_x0[dim], comm->old_cell_x1[dim]);
4367 fprintf(fplog, "New cell boundaries in direction %c: %8.3f %8.3f\n",
4369 comm->cell_x0[dim], comm->cell_x1[dim]);
4372 static void cg_move_error(FILE *fplog,
4374 gmx_int64_t step, int cg, int dim, int dir,
4375 gmx_bool bHaveLimitdAndCMOld, real limitd,
4376 rvec cm_old, rvec cm_new, real pos_d)
4380 print_cg_move(fplog, dd, step, cg, dim, dir,
4381 bHaveLimitdAndCMOld, limitd, cm_old, cm_new, pos_d);
4383 print_cg_move(stderr, dd, step, cg, dim, dir,
4384 bHaveLimitdAndCMOld, limitd, cm_old, cm_new, pos_d);
4386 "A charge group moved too far between two domain decomposition steps\n"
4387 "This usually means that your system is not well equilibrated");
4390 static void rotate_state_atom(t_state *state, int a)
4394 for (est = 0; est < estNR; est++)
4396 if (EST_DISTR(est) && (state->flags & (1<<est)))
4401 /* Rotate the complete state; for a rectangular box only */
4402 state->x[a][YY] = state->box[YY][YY] - state->x[a][YY];
4403 state->x[a][ZZ] = state->box[ZZ][ZZ] - state->x[a][ZZ];
4406 state->v[a][YY] = -state->v[a][YY];
4407 state->v[a][ZZ] = -state->v[a][ZZ];
4410 state->sd_X[a][YY] = -state->sd_X[a][YY];
4411 state->sd_X[a][ZZ] = -state->sd_X[a][ZZ];
4414 state->cg_p[a][YY] = -state->cg_p[a][YY];
4415 state->cg_p[a][ZZ] = -state->cg_p[a][ZZ];
4417 case estDISRE_INITF:
4418 case estDISRE_RM3TAV:
4419 case estORIRE_INITF:
4421 /* These are distances, so not affected by rotation */
4424 gmx_incons("Unknown state entry encountered in rotate_state_atom");
4430 static int *get_moved(gmx_domdec_comm_t *comm, int natoms)
4432 if (natoms > comm->moved_nalloc)
4434 /* Contents should be preserved here */
4435 comm->moved_nalloc = over_alloc_dd(natoms);
4436 srenew(comm->moved, comm->moved_nalloc);
4442 static void calc_cg_move(FILE *fplog, gmx_int64_t step,
4445 ivec tric_dir, matrix tcm,
4446 rvec cell_x0, rvec cell_x1,
4447 rvec limitd, rvec limit0, rvec limit1,
4449 int cg_start, int cg_end,
4454 int c, i, cg, k, k0, k1, d, dim, dim2, dir, d2, d3, d4, cell_d;
4455 int mc, cdd, nrcg, ncg_recv, nat_recv, nvs, nvr, nvec, vec;
4459 real inv_ncg, pos_d;
4462 npbcdim = dd->npbcdim;
4464 for (cg = cg_start; cg < cg_end; cg++)
4471 copy_rvec(state->x[k0], cm_new);
4478 for (k = k0; (k < k1); k++)
4480 rvec_inc(cm_new, state->x[k]);
4482 for (d = 0; (d < DIM); d++)
4484 cm_new[d] = inv_ncg*cm_new[d];
4489 /* Do pbc and check DD cell boundary crossings */
4490 for (d = DIM-1; d >= 0; d--)
4494 bScrew = (dd->bScrewPBC && d == XX);
4495 /* Determine the location of this cg in lattice coordinates */
4499 for (d2 = d+1; d2 < DIM; d2++)
4501 pos_d += cm_new[d2]*tcm[d2][d];
4504 /* Put the charge group in the triclinic unit-cell */
4505 if (pos_d >= cell_x1[d])
4507 if (pos_d >= limit1[d])
4509 cg_move_error(fplog, dd, step, cg, d, 1, TRUE, limitd[d],
4510 cg_cm[cg], cm_new, pos_d);
4513 if (dd->ci[d] == dd->nc[d] - 1)
4515 rvec_dec(cm_new, state->box[d]);
4518 cm_new[YY] = state->box[YY][YY] - cm_new[YY];
4519 cm_new[ZZ] = state->box[ZZ][ZZ] - cm_new[ZZ];
4521 for (k = k0; (k < k1); k++)
4523 rvec_dec(state->x[k], state->box[d]);
4526 rotate_state_atom(state, k);
4531 else if (pos_d < cell_x0[d])
4533 if (pos_d < limit0[d])
4535 cg_move_error(fplog, dd, step, cg, d, -1, TRUE, limitd[d],
4536 cg_cm[cg], cm_new, pos_d);
4541 rvec_inc(cm_new, state->box[d]);
4544 cm_new[YY] = state->box[YY][YY] - cm_new[YY];
4545 cm_new[ZZ] = state->box[ZZ][ZZ] - cm_new[ZZ];
4547 for (k = k0; (k < k1); k++)
4549 rvec_inc(state->x[k], state->box[d]);
4552 rotate_state_atom(state, k);
4558 else if (d < npbcdim)
4560 /* Put the charge group in the rectangular unit-cell */
4561 while (cm_new[d] >= state->box[d][d])
4563 rvec_dec(cm_new, state->box[d]);
4564 for (k = k0; (k < k1); k++)
4566 rvec_dec(state->x[k], state->box[d]);
4569 while (cm_new[d] < 0)
4571 rvec_inc(cm_new, state->box[d]);
4572 for (k = k0; (k < k1); k++)
4574 rvec_inc(state->x[k], state->box[d]);
4580 copy_rvec(cm_new, cg_cm[cg]);
4582 /* Determine where this cg should go */
4585 for (d = 0; d < dd->ndim; d++)
4590 flag |= DD_FLAG_FW(d);
4596 else if (dev[dim] == -1)
4598 flag |= DD_FLAG_BW(d);
4601 if (dd->nc[dim] > 2)
4612 /* Temporarily store the flag in move */
4613 move[cg] = mc + flag;
4617 static void dd_redistribute_cg(FILE *fplog, gmx_int64_t step,
4618 gmx_domdec_t *dd, ivec tric_dir,
4619 t_state *state, rvec **f,
4628 int ncg[DIM*2], nat[DIM*2];
4629 int c, i, cg, k, k0, k1, d, dim, dim2, dir, d2, d3, d4, cell_d;
4630 int mc, cdd, nrcg, ncg_recv, nat_recv, nvs, nvr, nvec, vec;
4631 int sbuf[2], rbuf[2];
4632 int home_pos_cg, home_pos_at, buf_pos;
4634 gmx_bool bV = FALSE, bSDX = FALSE, bCGP = FALSE;
4637 real inv_ncg, pos_d;
4639 rvec *cg_cm = NULL, cell_x0, cell_x1, limitd, limit0, limit1, cm_new;
4641 cginfo_mb_t *cginfo_mb;
4642 gmx_domdec_comm_t *comm;
4644 int nthread, thread;
4648 check_screw_box(state->box);
4652 if (fr->cutoff_scheme == ecutsGROUP)
4657 for (i = 0; i < estNR; i++)
4663 case estX: /* Always present */ break;
4664 case estV: bV = (state->flags & (1<<i)); break;
4665 case estSDX: bSDX = (state->flags & (1<<i)); break;
4666 case estCGP: bCGP = (state->flags & (1<<i)); break;
4669 case estDISRE_INITF:
4670 case estDISRE_RM3TAV:
4671 case estORIRE_INITF:
4673 /* No processing required */
4676 gmx_incons("Unknown state entry encountered in dd_redistribute_cg");
4681 if (dd->ncg_tot > comm->nalloc_int)
4683 comm->nalloc_int = over_alloc_dd(dd->ncg_tot);
4684 srenew(comm->buf_int, comm->nalloc_int);
4686 move = comm->buf_int;
4688 /* Clear the count */
4689 for (c = 0; c < dd->ndim*2; c++)
4695 npbcdim = dd->npbcdim;
4697 for (d = 0; (d < DIM); d++)
4699 limitd[d] = dd->comm->cellsize_min[d];
4700 if (d >= npbcdim && dd->ci[d] == 0)
4702 cell_x0[d] = -GMX_FLOAT_MAX;
4706 cell_x0[d] = comm->cell_x0[d];
4708 if (d >= npbcdim && dd->ci[d] == dd->nc[d] - 1)
4710 cell_x1[d] = GMX_FLOAT_MAX;
4714 cell_x1[d] = comm->cell_x1[d];
4718 limit0[d] = comm->old_cell_x0[d] - limitd[d];
4719 limit1[d] = comm->old_cell_x1[d] + limitd[d];
4723 /* We check after communication if a charge group moved
4724 * more than one cell. Set the pre-comm check limit to float_max.
4726 limit0[d] = -GMX_FLOAT_MAX;
4727 limit1[d] = GMX_FLOAT_MAX;
4731 make_tric_corr_matrix(npbcdim, state->box, tcm);
4733 cgindex = dd->cgindex;
4735 nthread = gmx_omp_nthreads_get(emntDomdec);
4737 /* Compute the center of geometry for all home charge groups
4738 * and put them in the box and determine where they should go.
4740 #pragma omp parallel for num_threads(nthread) schedule(static)
4741 for (thread = 0; thread < nthread; thread++)
4743 calc_cg_move(fplog, step, dd, state, tric_dir, tcm,
4744 cell_x0, cell_x1, limitd, limit0, limit1,
4746 ( thread *dd->ncg_home)/nthread,
4747 ((thread+1)*dd->ncg_home)/nthread,
4748 fr->cutoff_scheme == ecutsGROUP ? cg_cm : state->x,
4752 for (cg = 0; cg < dd->ncg_home; cg++)
4757 flag = mc & ~DD_FLAG_NRCG;
4758 mc = mc & DD_FLAG_NRCG;
4761 if (ncg[mc]+1 > comm->cggl_flag_nalloc[mc])
4763 comm->cggl_flag_nalloc[mc] = over_alloc_dd(ncg[mc]+1);
4764 srenew(comm->cggl_flag[mc], comm->cggl_flag_nalloc[mc]*DD_CGIBS);
4766 comm->cggl_flag[mc][ncg[mc]*DD_CGIBS ] = dd->index_gl[cg];
4767 /* We store the cg size in the lower 16 bits
4768 * and the place where the charge group should go
4769 * in the next 6 bits. This saves some communication volume.
4771 nrcg = cgindex[cg+1] - cgindex[cg];
4772 comm->cggl_flag[mc][ncg[mc]*DD_CGIBS+1] = nrcg | flag;
4778 inc_nrnb(nrnb, eNR_CGCM, dd->nat_home);
4779 inc_nrnb(nrnb, eNR_RESETX, dd->ncg_home);
4782 for (i = 0; i < dd->ndim*2; i++)
4784 *ncg_moved += ncg[i];
4801 /* Make sure the communication buffers are large enough */
4802 for (mc = 0; mc < dd->ndim*2; mc++)
4804 nvr = ncg[mc] + nat[mc]*nvec;
4805 if (nvr > comm->cgcm_state_nalloc[mc])
4807 comm->cgcm_state_nalloc[mc] = over_alloc_dd(nvr);
4808 srenew(comm->cgcm_state[mc], comm->cgcm_state_nalloc[mc]);
4812 switch (fr->cutoff_scheme)
4815 /* Recalculating cg_cm might be cheaper than communicating,
4816 * but that could give rise to rounding issues.
4819 compact_and_copy_vec_cg(dd->ncg_home, move, cgindex,
4820 nvec, cg_cm, comm, bCompact);
4823 /* Without charge groups we send the moved atom coordinates
4824 * over twice. This is so the code below can be used without
4825 * many conditionals for both for with and without charge groups.
4828 compact_and_copy_vec_cg(dd->ncg_home, move, cgindex,
4829 nvec, state->x, comm, FALSE);
4832 home_pos_cg -= *ncg_moved;
4836 gmx_incons("unimplemented");
4842 compact_and_copy_vec_at(dd->ncg_home, move, cgindex,
4843 nvec, vec++, state->x, comm, bCompact);
4846 compact_and_copy_vec_at(dd->ncg_home, move, cgindex,
4847 nvec, vec++, state->v, comm, bCompact);
4851 compact_and_copy_vec_at(dd->ncg_home, move, cgindex,
4852 nvec, vec++, state->sd_X, comm, bCompact);
4856 compact_and_copy_vec_at(dd->ncg_home, move, cgindex,
4857 nvec, vec++, state->cg_p, comm, bCompact);
4862 compact_ind(dd->ncg_home, move,
4863 dd->index_gl, dd->cgindex, dd->gatindex,
4864 dd->ga2la, comm->bLocalCG,
4869 if (fr->cutoff_scheme == ecutsVERLET)
4871 moved = get_moved(comm, dd->ncg_home);
4873 for (k = 0; k < dd->ncg_home; k++)
4880 moved = fr->ns.grid->cell_index;
4883 clear_and_mark_ind(dd->ncg_home, move,
4884 dd->index_gl, dd->cgindex, dd->gatindex,
4885 dd->ga2la, comm->bLocalCG,
4889 cginfo_mb = fr->cginfo_mb;
4891 *ncg_stay_home = home_pos_cg;
4892 for (d = 0; d < dd->ndim; d++)
4898 for (dir = 0; dir < (dd->nc[dim] == 2 ? 1 : 2); dir++)
4901 /* Communicate the cg and atom counts */
4906 fprintf(debug, "Sending ddim %d dir %d: ncg %d nat %d\n",
4907 d, dir, sbuf[0], sbuf[1]);
4909 dd_sendrecv_int(dd, d, dir, sbuf, 2, rbuf, 2);
4911 if ((ncg_recv+rbuf[0])*DD_CGIBS > comm->nalloc_int)
4913 comm->nalloc_int = over_alloc_dd((ncg_recv+rbuf[0])*DD_CGIBS);
4914 srenew(comm->buf_int, comm->nalloc_int);
4917 /* Communicate the charge group indices, sizes and flags */
4918 dd_sendrecv_int(dd, d, dir,
4919 comm->cggl_flag[cdd], sbuf[0]*DD_CGIBS,
4920 comm->buf_int+ncg_recv*DD_CGIBS, rbuf[0]*DD_CGIBS);
4922 nvs = ncg[cdd] + nat[cdd]*nvec;
4923 i = rbuf[0] + rbuf[1] *nvec;
4924 vec_rvec_check_alloc(&comm->vbuf, nvr+i);
4926 /* Communicate cgcm and state */
4927 dd_sendrecv_rvec(dd, d, dir,
4928 comm->cgcm_state[cdd], nvs,
4929 comm->vbuf.v+nvr, i);
4930 ncg_recv += rbuf[0];
4931 nat_recv += rbuf[1];
4935 /* Process the received charge groups */
4937 for (cg = 0; cg < ncg_recv; cg++)
4939 flag = comm->buf_int[cg*DD_CGIBS+1];
4941 if (dim >= npbcdim && dd->nc[dim] > 2)
4943 /* No pbc in this dim and more than one domain boundary.
4944 * We do a separate check if a charge group didn't move too far.
4946 if (((flag & DD_FLAG_FW(d)) &&
4947 comm->vbuf.v[buf_pos][dim] > cell_x1[dim]) ||
4948 ((flag & DD_FLAG_BW(d)) &&
4949 comm->vbuf.v[buf_pos][dim] < cell_x0[dim]))
4951 cg_move_error(fplog, dd, step, cg, dim,
4952 (flag & DD_FLAG_FW(d)) ? 1 : 0,
4954 comm->vbuf.v[buf_pos],
4955 comm->vbuf.v[buf_pos],
4956 comm->vbuf.v[buf_pos][dim]);
4963 /* Check which direction this cg should go */
4964 for (d2 = d+1; (d2 < dd->ndim && mc == -1); d2++)
4968 /* The cell boundaries for dimension d2 are not equal
4969 * for each cell row of the lower dimension(s),
4970 * therefore we might need to redetermine where
4971 * this cg should go.
4974 /* If this cg crosses the box boundary in dimension d2
4975 * we can use the communicated flag, so we do not
4976 * have to worry about pbc.
4978 if (!((dd->ci[dim2] == dd->nc[dim2]-1 &&
4979 (flag & DD_FLAG_FW(d2))) ||
4980 (dd->ci[dim2] == 0 &&
4981 (flag & DD_FLAG_BW(d2)))))
4983 /* Clear the two flags for this dimension */
4984 flag &= ~(DD_FLAG_FW(d2) | DD_FLAG_BW(d2));
4985 /* Determine the location of this cg
4986 * in lattice coordinates
4988 pos_d = comm->vbuf.v[buf_pos][dim2];
4991 for (d3 = dim2+1; d3 < DIM; d3++)
4994 comm->vbuf.v[buf_pos][d3]*tcm[d3][dim2];
4997 /* Check of we are not at the box edge.
4998 * pbc is only handled in the first step above,
4999 * but this check could move over pbc while
5000 * the first step did not due to different rounding.
5002 if (pos_d >= cell_x1[dim2] &&
5003 dd->ci[dim2] != dd->nc[dim2]-1)
5005 flag |= DD_FLAG_FW(d2);
5007 else if (pos_d < cell_x0[dim2] &&
5010 flag |= DD_FLAG_BW(d2);
5012 comm->buf_int[cg*DD_CGIBS+1] = flag;
5015 /* Set to which neighboring cell this cg should go */
5016 if (flag & DD_FLAG_FW(d2))
5020 else if (flag & DD_FLAG_BW(d2))
5022 if (dd->nc[dd->dim[d2]] > 2)
5034 nrcg = flag & DD_FLAG_NRCG;
5037 if (home_pos_cg+1 > dd->cg_nalloc)
5039 dd->cg_nalloc = over_alloc_dd(home_pos_cg+1);
5040 srenew(dd->index_gl, dd->cg_nalloc);
5041 srenew(dd->cgindex, dd->cg_nalloc+1);
5043 /* Set the global charge group index and size */
5044 dd->index_gl[home_pos_cg] = comm->buf_int[cg*DD_CGIBS];
5045 dd->cgindex[home_pos_cg+1] = dd->cgindex[home_pos_cg] + nrcg;
5046 /* Copy the state from the buffer */
5047 dd_check_alloc_ncg(fr, state, f, home_pos_cg+1);
5048 if (fr->cutoff_scheme == ecutsGROUP)
5051 copy_rvec(comm->vbuf.v[buf_pos], cg_cm[home_pos_cg]);
5055 /* Set the cginfo */
5056 fr->cginfo[home_pos_cg] = ddcginfo(cginfo_mb,
5057 dd->index_gl[home_pos_cg]);
5060 comm->bLocalCG[dd->index_gl[home_pos_cg]] = TRUE;
5063 if (home_pos_at+nrcg > state->nalloc)
5065 dd_realloc_state(state, f, home_pos_at+nrcg);
5067 for (i = 0; i < nrcg; i++)
5069 copy_rvec(comm->vbuf.v[buf_pos++],
5070 state->x[home_pos_at+i]);
5074 for (i = 0; i < nrcg; i++)
5076 copy_rvec(comm->vbuf.v[buf_pos++],
5077 state->v[home_pos_at+i]);
5082 for (i = 0; i < nrcg; i++)
5084 copy_rvec(comm->vbuf.v[buf_pos++],
5085 state->sd_X[home_pos_at+i]);
5090 for (i = 0; i < nrcg; i++)
5092 copy_rvec(comm->vbuf.v[buf_pos++],
5093 state->cg_p[home_pos_at+i]);
5097 home_pos_at += nrcg;
5101 /* Reallocate the buffers if necessary */
5102 if (ncg[mc]+1 > comm->cggl_flag_nalloc[mc])
5104 comm->cggl_flag_nalloc[mc] = over_alloc_dd(ncg[mc]+1);
5105 srenew(comm->cggl_flag[mc], comm->cggl_flag_nalloc[mc]*DD_CGIBS);
5107 nvr = ncg[mc] + nat[mc]*nvec;
5108 if (nvr + 1 + nrcg*nvec > comm->cgcm_state_nalloc[mc])
5110 comm->cgcm_state_nalloc[mc] = over_alloc_dd(nvr + 1 + nrcg*nvec);
5111 srenew(comm->cgcm_state[mc], comm->cgcm_state_nalloc[mc]);
5113 /* Copy from the receive to the send buffers */
5114 memcpy(comm->cggl_flag[mc] + ncg[mc]*DD_CGIBS,
5115 comm->buf_int + cg*DD_CGIBS,
5116 DD_CGIBS*sizeof(int));
5117 memcpy(comm->cgcm_state[mc][nvr],
5118 comm->vbuf.v[buf_pos],
5119 (1+nrcg*nvec)*sizeof(rvec));
5120 buf_pos += 1 + nrcg*nvec;
5127 /* With sorting (!bCompact) the indices are now only partially up to date
5128 * and ncg_home and nat_home are not the real count, since there are
5129 * "holes" in the arrays for the charge groups that moved to neighbors.
5131 if (fr->cutoff_scheme == ecutsVERLET)
5133 moved = get_moved(comm, home_pos_cg);
5135 for (i = dd->ncg_home; i < home_pos_cg; i++)
5140 dd->ncg_home = home_pos_cg;
5141 dd->nat_home = home_pos_at;
5146 "Finished repartitioning: cgs moved out %d, new home %d\n",
5147 *ncg_moved, dd->ncg_home-*ncg_moved);
5152 void dd_cycles_add(gmx_domdec_t *dd, float cycles, int ddCycl)
5154 dd->comm->cycl[ddCycl] += cycles;
5155 dd->comm->cycl_n[ddCycl]++;
5156 if (cycles > dd->comm->cycl_max[ddCycl])
5158 dd->comm->cycl_max[ddCycl] = cycles;
5162 static double force_flop_count(t_nrnb *nrnb)
5169 for (i = 0; i < eNR_NBKERNEL_FREE_ENERGY; i++)
5171 /* To get closer to the real timings, we half the count
5172 * for the normal loops and again half it for water loops.
5175 if (strstr(name, "W3") != NULL || strstr(name, "W4") != NULL)
5177 sum += nrnb->n[i]*0.25*cost_nrnb(i);
5181 sum += nrnb->n[i]*0.50*cost_nrnb(i);
5184 for (i = eNR_NBKERNEL_FREE_ENERGY; i <= eNR_NB14; i++)
5187 if (strstr(name, "W3") != NULL || strstr(name, "W4") != NULL)
5189 sum += nrnb->n[i]*cost_nrnb(i);
5192 for (i = eNR_BONDS; i <= eNR_WALLS; i++)
5194 sum += nrnb->n[i]*cost_nrnb(i);
5200 void dd_force_flop_start(gmx_domdec_t *dd, t_nrnb *nrnb)
5202 if (dd->comm->eFlop)
5204 dd->comm->flop -= force_flop_count(nrnb);
5207 void dd_force_flop_stop(gmx_domdec_t *dd, t_nrnb *nrnb)
5209 if (dd->comm->eFlop)
5211 dd->comm->flop += force_flop_count(nrnb);
5216 static void clear_dd_cycle_counts(gmx_domdec_t *dd)
5220 for (i = 0; i < ddCyclNr; i++)
5222 dd->comm->cycl[i] = 0;
5223 dd->comm->cycl_n[i] = 0;
5224 dd->comm->cycl_max[i] = 0;
5227 dd->comm->flop_n = 0;
5230 static void get_load_distribution(gmx_domdec_t *dd, gmx_wallcycle_t wcycle)
5232 gmx_domdec_comm_t *comm;
5233 gmx_domdec_load_t *load;
5234 gmx_domdec_root_t *root = NULL;
5235 int d, dim, cid, i, pos;
5236 float cell_frac = 0, sbuf[DD_NLOAD_MAX];
5241 fprintf(debug, "get_load_distribution start\n");
5244 wallcycle_start(wcycle, ewcDDCOMMLOAD);
5248 bSepPME = (dd->pme_nodeid >= 0);
5250 for (d = dd->ndim-1; d >= 0; d--)
5253 /* Check if we participate in the communication in this dimension */
5254 if (d == dd->ndim-1 ||
5255 (dd->ci[dd->dim[d+1]] == 0 && dd->ci[dd->dim[dd->ndim-1]] == 0))
5257 load = &comm->load[d];
5260 cell_frac = comm->cell_f1[d] - comm->cell_f0[d];
5263 if (d == dd->ndim-1)
5265 sbuf[pos++] = dd_force_load(comm);
5266 sbuf[pos++] = sbuf[0];
5269 sbuf[pos++] = sbuf[0];
5270 sbuf[pos++] = cell_frac;
5273 sbuf[pos++] = comm->cell_f_max0[d];
5274 sbuf[pos++] = comm->cell_f_min1[d];
5279 sbuf[pos++] = comm->cycl[ddCyclPPduringPME];
5280 sbuf[pos++] = comm->cycl[ddCyclPME];
5285 sbuf[pos++] = comm->load[d+1].sum;
5286 sbuf[pos++] = comm->load[d+1].max;
5289 sbuf[pos++] = comm->load[d+1].sum_m;
5290 sbuf[pos++] = comm->load[d+1].cvol_min*cell_frac;
5291 sbuf[pos++] = comm->load[d+1].flags;
5294 sbuf[pos++] = comm->cell_f_max0[d];
5295 sbuf[pos++] = comm->cell_f_min1[d];
5300 sbuf[pos++] = comm->load[d+1].mdf;
5301 sbuf[pos++] = comm->load[d+1].pme;
5305 /* Communicate a row in DD direction d.
5306 * The communicators are setup such that the root always has rank 0.
5309 MPI_Gather(sbuf, load->nload*sizeof(float), MPI_BYTE,
5310 load->load, load->nload*sizeof(float), MPI_BYTE,
5311 0, comm->mpi_comm_load[d]);
5313 if (dd->ci[dim] == dd->master_ci[dim])
5315 /* We are the root, process this row */
5316 if (comm->bDynLoadBal)
5318 root = comm->root[d];
5328 for (i = 0; i < dd->nc[dim]; i++)
5330 load->sum += load->load[pos++];
5331 load->max = max(load->max, load->load[pos]);
5337 /* This direction could not be load balanced properly,
5338 * therefore we need to use the maximum iso the average load.
5340 load->sum_m = max(load->sum_m, load->load[pos]);
5344 load->sum_m += load->load[pos];
5347 load->cvol_min = min(load->cvol_min, load->load[pos]);
5351 load->flags = (int)(load->load[pos++] + 0.5);
5355 root->cell_f_max0[i] = load->load[pos++];
5356 root->cell_f_min1[i] = load->load[pos++];
5361 load->mdf = max(load->mdf, load->load[pos]);
5363 load->pme = max(load->pme, load->load[pos]);
5367 if (comm->bDynLoadBal && root->bLimited)
5369 load->sum_m *= dd->nc[dim];
5370 load->flags |= (1<<d);
5378 comm->nload += dd_load_count(comm);
5379 comm->load_step += comm->cycl[ddCyclStep];
5380 comm->load_sum += comm->load[0].sum;
5381 comm->load_max += comm->load[0].max;
5382 if (comm->bDynLoadBal)
5384 for (d = 0; d < dd->ndim; d++)
5386 if (comm->load[0].flags & (1<<d))
5388 comm->load_lim[d]++;
5394 comm->load_mdf += comm->load[0].mdf;
5395 comm->load_pme += comm->load[0].pme;
5399 wallcycle_stop(wcycle, ewcDDCOMMLOAD);
5403 fprintf(debug, "get_load_distribution finished\n");
5407 static float dd_force_imb_perf_loss(gmx_domdec_t *dd)
5409 /* Return the relative performance loss on the total run time
5410 * due to the force calculation load imbalance.
5412 if (dd->comm->nload > 0)
5415 (dd->comm->load_max*dd->nnodes - dd->comm->load_sum)/
5416 (dd->comm->load_step*dd->nnodes);
5424 static void print_dd_load_av(FILE *fplog, gmx_domdec_t *dd)
5427 int npp, npme, nnodes, d, limp;
5428 float imbal, pme_f_ratio, lossf, lossp = 0;
5430 gmx_domdec_comm_t *comm;
5433 if (DDMASTER(dd) && comm->nload > 0)
5436 npme = (dd->pme_nodeid >= 0) ? comm->npmenodes : 0;
5437 nnodes = npp + npme;
5438 imbal = comm->load_max*npp/comm->load_sum - 1;
5439 lossf = dd_force_imb_perf_loss(dd);
5440 sprintf(buf, " Average load imbalance: %.1f %%\n", imbal*100);
5441 fprintf(fplog, "%s", buf);
5442 fprintf(stderr, "\n");
5443 fprintf(stderr, "%s", buf);
5444 sprintf(buf, " Part of the total run time spent waiting due to load imbalance: %.1f %%\n", lossf*100);
5445 fprintf(fplog, "%s", buf);
5446 fprintf(stderr, "%s", buf);
5448 if (comm->bDynLoadBal)
5450 sprintf(buf, " Steps where the load balancing was limited by -rdd, -rcon and/or -dds:");
5451 for (d = 0; d < dd->ndim; d++)
5453 limp = (200*comm->load_lim[d]+1)/(2*comm->nload);
5454 sprintf(buf+strlen(buf), " %c %d %%", dim2char(dd->dim[d]), limp);
5460 sprintf(buf+strlen(buf), "\n");
5461 fprintf(fplog, "%s", buf);
5462 fprintf(stderr, "%s", buf);
5466 pme_f_ratio = comm->load_pme/comm->load_mdf;
5467 lossp = (comm->load_pme -comm->load_mdf)/comm->load_step;
5470 lossp *= (float)npme/(float)nnodes;
5474 lossp *= (float)npp/(float)nnodes;
5476 sprintf(buf, " Average PME mesh/force load: %5.3f\n", pme_f_ratio);
5477 fprintf(fplog, "%s", buf);
5478 fprintf(stderr, "%s", buf);
5479 sprintf(buf, " Part of the total run time spent waiting due to PP/PME imbalance: %.1f %%\n", fabs(lossp)*100);
5480 fprintf(fplog, "%s", buf);
5481 fprintf(stderr, "%s", buf);
5483 fprintf(fplog, "\n");
5484 fprintf(stderr, "\n");
5486 if (lossf >= DD_PERF_LOSS_WARN)
5489 "NOTE: %.1f %% of the available CPU time was lost due to load imbalance\n"
5490 " in the domain decomposition.\n", lossf*100);
5491 if (!comm->bDynLoadBal)
5493 sprintf(buf+strlen(buf), " You might want to use dynamic load balancing (option -dlb.)\n");
5497 sprintf(buf+strlen(buf), " You might want to decrease the cell size limit (options -rdd, -rcon and/or -dds).\n");
5499 fprintf(fplog, "%s\n", buf);
5500 fprintf(stderr, "%s\n", buf);
5502 if (npme > 0 && fabs(lossp) >= DD_PERF_LOSS_WARN)
5505 "NOTE: %.1f %% performance was lost because the PME ranks\n"
5506 " had %s work to do than the PP ranks.\n"
5507 " You might want to %s the number of PME ranks\n"
5508 " or %s the cut-off and the grid spacing.\n",
5510 (lossp < 0) ? "less" : "more",
5511 (lossp < 0) ? "decrease" : "increase",
5512 (lossp < 0) ? "decrease" : "increase");
5513 fprintf(fplog, "%s\n", buf);
5514 fprintf(stderr, "%s\n", buf);
5519 static float dd_vol_min(gmx_domdec_t *dd)
5521 return dd->comm->load[0].cvol_min*dd->nnodes;
5524 static gmx_bool dd_load_flags(gmx_domdec_t *dd)
5526 return dd->comm->load[0].flags;
5529 static float dd_f_imbal(gmx_domdec_t *dd)
5531 return dd->comm->load[0].max*dd->nnodes/dd->comm->load[0].sum - 1;
5534 float dd_pme_f_ratio(gmx_domdec_t *dd)
5536 if (dd->comm->cycl_n[ddCyclPME] > 0)
5538 return dd->comm->load[0].pme/dd->comm->load[0].mdf;
5546 static void dd_print_load(FILE *fplog, gmx_domdec_t *dd, gmx_int64_t step)
5551 flags = dd_load_flags(dd);
5555 "DD load balancing is limited by minimum cell size in dimension");
5556 for (d = 0; d < dd->ndim; d++)
5560 fprintf(fplog, " %c", dim2char(dd->dim[d]));
5563 fprintf(fplog, "\n");
5565 fprintf(fplog, "DD step %s", gmx_step_str(step, buf));
5566 if (dd->comm->bDynLoadBal)
5568 fprintf(fplog, " vol min/aver %5.3f%c",
5569 dd_vol_min(dd), flags ? '!' : ' ');
5571 fprintf(fplog, " load imb.: force %4.1f%%", dd_f_imbal(dd)*100);
5572 if (dd->comm->cycl_n[ddCyclPME])
5574 fprintf(fplog, " pme mesh/force %5.3f", dd_pme_f_ratio(dd));
5576 fprintf(fplog, "\n\n");
5579 static void dd_print_load_verbose(gmx_domdec_t *dd)
5581 if (dd->comm->bDynLoadBal)
5583 fprintf(stderr, "vol %4.2f%c ",
5584 dd_vol_min(dd), dd_load_flags(dd) ? '!' : ' ');
5586 fprintf(stderr, "imb F %2d%% ", (int)(dd_f_imbal(dd)*100+0.5));
5587 if (dd->comm->cycl_n[ddCyclPME])
5589 fprintf(stderr, "pme/F %4.2f ", dd_pme_f_ratio(dd));
5594 static void make_load_communicator(gmx_domdec_t *dd, int dim_ind, ivec loc)
5599 gmx_domdec_root_t *root;
5600 gmx_bool bPartOfGroup = FALSE;
5602 dim = dd->dim[dim_ind];
5603 copy_ivec(loc, loc_c);
5604 for (i = 0; i < dd->nc[dim]; i++)
5607 rank = dd_index(dd->nc, loc_c);
5608 if (rank == dd->rank)
5610 /* This process is part of the group */
5611 bPartOfGroup = TRUE;
5614 MPI_Comm_split(dd->mpi_comm_all, bPartOfGroup ? 0 : MPI_UNDEFINED, dd->rank,
5618 dd->comm->mpi_comm_load[dim_ind] = c_row;
5619 if (dd->comm->eDLB != edlbNO)
5621 if (dd->ci[dim] == dd->master_ci[dim])
5623 /* This is the root process of this row */
5624 snew(dd->comm->root[dim_ind], 1);
5625 root = dd->comm->root[dim_ind];
5626 snew(root->cell_f, DD_CELL_F_SIZE(dd, dim_ind));
5627 snew(root->old_cell_f, dd->nc[dim]+1);
5628 snew(root->bCellMin, dd->nc[dim]);
5631 snew(root->cell_f_max0, dd->nc[dim]);
5632 snew(root->cell_f_min1, dd->nc[dim]);
5633 snew(root->bound_min, dd->nc[dim]);
5634 snew(root->bound_max, dd->nc[dim]);
5636 snew(root->buf_ncd, dd->nc[dim]);
5640 /* This is not a root process, we only need to receive cell_f */
5641 snew(dd->comm->cell_f_row, DD_CELL_F_SIZE(dd, dim_ind));
5644 if (dd->ci[dim] == dd->master_ci[dim])
5646 snew(dd->comm->load[dim_ind].load, dd->nc[dim]*DD_NLOAD_MAX);
5652 void dd_setup_dlb_resource_sharing(t_commrec gmx_unused *cr,
5653 const gmx_hw_info_t gmx_unused *hwinfo,
5654 const gmx_hw_opt_t gmx_unused *hw_opt)
5657 int physicalnode_id_hash;
5660 MPI_Comm mpi_comm_pp_physicalnode;
5662 if (!(cr->duty & DUTY_PP) ||
5663 hw_opt->gpu_opt.ncuda_dev_use == 0)
5665 /* Only PP nodes (currently) use GPUs.
5666 * If we don't have GPUs, there are no resources to share.
5671 physicalnode_id_hash = gmx_physicalnode_id_hash();
5673 gpu_id = get_gpu_device_id(&hwinfo->gpu_info, &hw_opt->gpu_opt, cr->rank_pp_intranode);
5679 fprintf(debug, "dd_setup_dd_dlb_gpu_sharing:\n");
5680 fprintf(debug, "DD PP rank %d physical node hash %d gpu_id %d\n",
5681 dd->rank, physicalnode_id_hash, gpu_id);
5683 /* Split the PP communicator over the physical nodes */
5684 /* TODO: See if we should store this (before), as it's also used for
5685 * for the nodecomm summution.
5687 MPI_Comm_split(dd->mpi_comm_all, physicalnode_id_hash, dd->rank,
5688 &mpi_comm_pp_physicalnode);
5689 MPI_Comm_split(mpi_comm_pp_physicalnode, gpu_id, dd->rank,
5690 &dd->comm->mpi_comm_gpu_shared);
5691 MPI_Comm_free(&mpi_comm_pp_physicalnode);
5692 MPI_Comm_size(dd->comm->mpi_comm_gpu_shared, &dd->comm->nrank_gpu_shared);
5696 fprintf(debug, "nrank_gpu_shared %d\n", dd->comm->nrank_gpu_shared);
5699 /* Note that some ranks could share a GPU, while others don't */
5701 if (dd->comm->nrank_gpu_shared == 1)
5703 MPI_Comm_free(&dd->comm->mpi_comm_gpu_shared);
5708 static void make_load_communicators(gmx_domdec_t gmx_unused *dd)
5711 int dim0, dim1, i, j;
5716 fprintf(debug, "Making load communicators\n");
5719 snew(dd->comm->load, dd->ndim);
5720 snew(dd->comm->mpi_comm_load, dd->ndim);
5723 make_load_communicator(dd, 0, loc);
5727 for (i = 0; i < dd->nc[dim0]; i++)
5730 make_load_communicator(dd, 1, loc);
5736 for (i = 0; i < dd->nc[dim0]; i++)
5740 for (j = 0; j < dd->nc[dim1]; j++)
5743 make_load_communicator(dd, 2, loc);
5750 fprintf(debug, "Finished making load communicators\n");
5755 void setup_dd_grid(FILE *fplog, gmx_domdec_t *dd)
5758 int d, dim, i, j, m;
5761 ivec dd_zp[DD_MAXIZONE];
5762 gmx_domdec_zones_t *zones;
5763 gmx_domdec_ns_ranges_t *izone;
5765 for (d = 0; d < dd->ndim; d++)
5768 copy_ivec(dd->ci, tmp);
5769 tmp[dim] = (tmp[dim] + 1) % dd->nc[dim];
5770 dd->neighbor[d][0] = ddcoord2ddnodeid(dd, tmp);
5771 copy_ivec(dd->ci, tmp);
5772 tmp[dim] = (tmp[dim] - 1 + dd->nc[dim]) % dd->nc[dim];
5773 dd->neighbor[d][1] = ddcoord2ddnodeid(dd, tmp);
5776 fprintf(debug, "DD rank %d neighbor ranks in dir %d are + %d - %d\n",
5779 dd->neighbor[d][1]);
5785 fprintf(fplog, "\nMaking %dD domain decomposition grid %d x %d x %d, home cell index %d %d %d\n\n",
5787 dd->nc[XX], dd->nc[YY], dd->nc[ZZ],
5788 dd->ci[XX], dd->ci[YY], dd->ci[ZZ]);
5795 for (i = 0; i < nzonep; i++)
5797 copy_ivec(dd_zp3[i], dd_zp[i]);
5803 for (i = 0; i < nzonep; i++)
5805 copy_ivec(dd_zp2[i], dd_zp[i]);
5811 for (i = 0; i < nzonep; i++)
5813 copy_ivec(dd_zp1[i], dd_zp[i]);
5817 gmx_fatal(FARGS, "Can only do 1, 2 or 3D domain decomposition");
5822 zones = &dd->comm->zones;
5824 for (i = 0; i < nzone; i++)
5827 clear_ivec(zones->shift[i]);
5828 for (d = 0; d < dd->ndim; d++)
5830 zones->shift[i][dd->dim[d]] = dd_zo[i][m++];
5835 for (i = 0; i < nzone; i++)
5837 for (d = 0; d < DIM; d++)
5839 s[d] = dd->ci[d] - zones->shift[i][d];
5844 else if (s[d] >= dd->nc[d])
5850 zones->nizone = nzonep;
5851 for (i = 0; i < zones->nizone; i++)
5853 if (dd_zp[i][0] != i)
5855 gmx_fatal(FARGS, "Internal inconsistency in the dd grid setup");
5857 izone = &zones->izone[i];
5858 izone->j0 = dd_zp[i][1];
5859 izone->j1 = dd_zp[i][2];
5860 for (dim = 0; dim < DIM; dim++)
5862 if (dd->nc[dim] == 1)
5864 /* All shifts should be allowed */
5865 izone->shift0[dim] = -1;
5866 izone->shift1[dim] = 1;
5871 izone->shift0[d] = 0;
5872 izone->shift1[d] = 0;
5873 for(j=izone->j0; j<izone->j1; j++) {
5874 if (dd->shift[j][d] > dd->shift[i][d])
5875 izone->shift0[d] = -1;
5876 if (dd->shift[j][d] < dd->shift[i][d])
5877 izone->shift1[d] = 1;
5883 /* Assume the shift are not more than 1 cell */
5884 izone->shift0[dim] = 1;
5885 izone->shift1[dim] = -1;
5886 for (j = izone->j0; j < izone->j1; j++)
5888 shift_diff = zones->shift[j][dim] - zones->shift[i][dim];
5889 if (shift_diff < izone->shift0[dim])
5891 izone->shift0[dim] = shift_diff;
5893 if (shift_diff > izone->shift1[dim])
5895 izone->shift1[dim] = shift_diff;
5902 if (dd->comm->eDLB != edlbNO)
5904 snew(dd->comm->root, dd->ndim);
5907 if (dd->comm->bRecordLoad)
5909 make_load_communicators(dd);
5913 static void make_pp_communicator(FILE *fplog, t_commrec *cr, int gmx_unused reorder)
5916 gmx_domdec_comm_t *comm;
5927 if (comm->bCartesianPP)
5929 /* Set up cartesian communication for the particle-particle part */
5932 fprintf(fplog, "Will use a Cartesian communicator: %d x %d x %d\n",
5933 dd->nc[XX], dd->nc[YY], dd->nc[ZZ]);
5936 for (i = 0; i < DIM; i++)
5940 MPI_Cart_create(cr->mpi_comm_mygroup, DIM, dd->nc, periods, reorder,
5942 /* We overwrite the old communicator with the new cartesian one */
5943 cr->mpi_comm_mygroup = comm_cart;
5946 dd->mpi_comm_all = cr->mpi_comm_mygroup;
5947 MPI_Comm_rank(dd->mpi_comm_all, &dd->rank);
5949 if (comm->bCartesianPP_PME)
5951 /* Since we want to use the original cartesian setup for sim,
5952 * and not the one after split, we need to make an index.
5954 snew(comm->ddindex2ddnodeid, dd->nnodes);
5955 comm->ddindex2ddnodeid[dd_index(dd->nc, dd->ci)] = dd->rank;
5956 gmx_sumi(dd->nnodes, comm->ddindex2ddnodeid, cr);
5957 /* Get the rank of the DD master,
5958 * above we made sure that the master node is a PP node.
5968 MPI_Allreduce(&rank, &dd->masterrank, 1, MPI_INT, MPI_SUM, dd->mpi_comm_all);
5970 else if (comm->bCartesianPP)
5972 if (cr->npmenodes == 0)
5974 /* The PP communicator is also
5975 * the communicator for this simulation
5977 cr->mpi_comm_mysim = cr->mpi_comm_mygroup;
5979 cr->nodeid = dd->rank;
5981 MPI_Cart_coords(dd->mpi_comm_all, dd->rank, DIM, dd->ci);
5983 /* We need to make an index to go from the coordinates
5984 * to the nodeid of this simulation.
5986 snew(comm->ddindex2simnodeid, dd->nnodes);
5987 snew(buf, dd->nnodes);
5988 if (cr->duty & DUTY_PP)
5990 buf[dd_index(dd->nc, dd->ci)] = cr->sim_nodeid;
5992 /* Communicate the ddindex to simulation nodeid index */
5993 MPI_Allreduce(buf, comm->ddindex2simnodeid, dd->nnodes, MPI_INT, MPI_SUM,
5994 cr->mpi_comm_mysim);
5997 /* Determine the master coordinates and rank.
5998 * The DD master should be the same node as the master of this sim.
6000 for (i = 0; i < dd->nnodes; i++)
6002 if (comm->ddindex2simnodeid[i] == 0)
6004 ddindex2xyz(dd->nc, i, dd->master_ci);
6005 MPI_Cart_rank(dd->mpi_comm_all, dd->master_ci, &dd->masterrank);
6010 fprintf(debug, "The master rank is %d\n", dd->masterrank);
6015 /* No Cartesian communicators */
6016 /* We use the rank in dd->comm->all as DD index */
6017 ddindex2xyz(dd->nc, dd->rank, dd->ci);
6018 /* The simulation master nodeid is 0, so the DD master rank is also 0 */
6020 clear_ivec(dd->master_ci);
6027 "Domain decomposition rank %d, coordinates %d %d %d\n\n",
6028 dd->rank, dd->ci[XX], dd->ci[YY], dd->ci[ZZ]);
6033 "Domain decomposition rank %d, coordinates %d %d %d\n\n",
6034 dd->rank, dd->ci[XX], dd->ci[YY], dd->ci[ZZ]);
6038 static void receive_ddindex2simnodeid(t_commrec *cr)
6042 gmx_domdec_comm_t *comm;
6049 if (!comm->bCartesianPP_PME && comm->bCartesianPP)
6051 snew(comm->ddindex2simnodeid, dd->nnodes);
6052 snew(buf, dd->nnodes);
6053 if (cr->duty & DUTY_PP)
6055 buf[dd_index(dd->nc, dd->ci)] = cr->sim_nodeid;
6058 /* Communicate the ddindex to simulation nodeid index */
6059 MPI_Allreduce(buf, comm->ddindex2simnodeid, dd->nnodes, MPI_INT, MPI_SUM,
6060 cr->mpi_comm_mysim);
6067 static gmx_domdec_master_t *init_gmx_domdec_master_t(gmx_domdec_t *dd,
6068 int ncg, int natoms)
6070 gmx_domdec_master_t *ma;
6075 snew(ma->ncg, dd->nnodes);
6076 snew(ma->index, dd->nnodes+1);
6078 snew(ma->nat, dd->nnodes);
6079 snew(ma->ibuf, dd->nnodes*2);
6080 snew(ma->cell_x, DIM);
6081 for (i = 0; i < DIM; i++)
6083 snew(ma->cell_x[i], dd->nc[i]+1);
6086 if (dd->nnodes <= GMX_DD_NNODES_SENDRECV)
6092 snew(ma->vbuf, natoms);
6098 static void split_communicator(FILE *fplog, t_commrec *cr, int gmx_unused dd_node_order,
6099 int gmx_unused reorder)
6102 gmx_domdec_comm_t *comm;
6113 if (comm->bCartesianPP)
6115 for (i = 1; i < DIM; i++)
6117 bDiv[i] = ((cr->npmenodes*dd->nc[i]) % (dd->nnodes) == 0);
6119 if (bDiv[YY] || bDiv[ZZ])
6121 comm->bCartesianPP_PME = TRUE;
6122 /* If we have 2D PME decomposition, which is always in x+y,
6123 * we stack the PME only nodes in z.
6124 * Otherwise we choose the direction that provides the thinnest slab
6125 * of PME only nodes as this will have the least effect
6126 * on the PP communication.
6127 * But for the PME communication the opposite might be better.
6129 if (bDiv[ZZ] && (comm->npmenodes_y > 1 ||
6131 dd->nc[YY] > dd->nc[ZZ]))
6133 comm->cartpmedim = ZZ;
6137 comm->cartpmedim = YY;
6139 comm->ntot[comm->cartpmedim]
6140 += (cr->npmenodes*dd->nc[comm->cartpmedim])/dd->nnodes;
6144 fprintf(fplog, "Number of PME-only ranks (%d) is not a multiple of nx*ny (%d*%d) or nx*nz (%d*%d)\n", cr->npmenodes, dd->nc[XX], dd->nc[YY], dd->nc[XX], dd->nc[ZZ]);
6146 "Will not use a Cartesian communicator for PP <-> PME\n\n");
6151 if (comm->bCartesianPP_PME)
6155 fprintf(fplog, "Will use a Cartesian communicator for PP <-> PME: %d x %d x %d\n", comm->ntot[XX], comm->ntot[YY], comm->ntot[ZZ]);
6158 for (i = 0; i < DIM; i++)
6162 MPI_Cart_create(cr->mpi_comm_mysim, DIM, comm->ntot, periods, reorder,
6165 MPI_Comm_rank(comm_cart, &rank);
6166 if (MASTERNODE(cr) && rank != 0)
6168 gmx_fatal(FARGS, "MPI rank 0 was renumbered by MPI_Cart_create, we do not allow this");
6171 /* With this assigment we loose the link to the original communicator
6172 * which will usually be MPI_COMM_WORLD, unless have multisim.
6174 cr->mpi_comm_mysim = comm_cart;
6175 cr->sim_nodeid = rank;
6177 MPI_Cart_coords(cr->mpi_comm_mysim, cr->sim_nodeid, DIM, dd->ci);
6181 fprintf(fplog, "Cartesian rank %d, coordinates %d %d %d\n\n",
6182 cr->sim_nodeid, dd->ci[XX], dd->ci[YY], dd->ci[ZZ]);
6185 if (dd->ci[comm->cartpmedim] < dd->nc[comm->cartpmedim])
6189 if (cr->npmenodes == 0 ||
6190 dd->ci[comm->cartpmedim] >= dd->nc[comm->cartpmedim])
6192 cr->duty = DUTY_PME;
6195 /* Split the sim communicator into PP and PME only nodes */
6196 MPI_Comm_split(cr->mpi_comm_mysim,
6198 dd_index(comm->ntot, dd->ci),
6199 &cr->mpi_comm_mygroup);
6203 switch (dd_node_order)
6208 fprintf(fplog, "Order of the ranks: PP first, PME last\n");
6211 case ddnoINTERLEAVE:
6212 /* Interleave the PP-only and PME-only nodes,
6213 * as on clusters with dual-core machines this will double
6214 * the communication bandwidth of the PME processes
6215 * and thus speed up the PP <-> PME and inter PME communication.
6219 fprintf(fplog, "Interleaving PP and PME ranks\n");
6221 comm->pmenodes = dd_pmenodes(cr);
6226 gmx_fatal(FARGS, "Unknown dd_node_order=%d", dd_node_order);
6229 if (dd_simnode2pmenode(cr, cr->sim_nodeid) == -1)
6231 cr->duty = DUTY_PME;
6238 /* Split the sim communicator into PP and PME only nodes */
6239 MPI_Comm_split(cr->mpi_comm_mysim,
6242 &cr->mpi_comm_mygroup);
6243 MPI_Comm_rank(cr->mpi_comm_mygroup, &cr->nodeid);
6249 fprintf(fplog, "This rank does only %s work.\n\n",
6250 (cr->duty & DUTY_PP) ? "particle-particle" : "PME-mesh");
6254 void make_dd_communicators(FILE *fplog, t_commrec *cr, int dd_node_order)
6257 gmx_domdec_comm_t *comm;
6263 copy_ivec(dd->nc, comm->ntot);
6265 comm->bCartesianPP = (dd_node_order == ddnoCARTESIAN);
6266 comm->bCartesianPP_PME = FALSE;
6268 /* Reorder the nodes by default. This might change the MPI ranks.
6269 * Real reordering is only supported on very few architectures,
6270 * Blue Gene is one of them.
6272 CartReorder = (getenv("GMX_NO_CART_REORDER") == NULL);
6274 if (cr->npmenodes > 0)
6276 /* Split the communicator into a PP and PME part */
6277 split_communicator(fplog, cr, dd_node_order, CartReorder);
6278 if (comm->bCartesianPP_PME)
6280 /* We (possibly) reordered the nodes in split_communicator,
6281 * so it is no longer required in make_pp_communicator.
6283 CartReorder = FALSE;
6288 /* All nodes do PP and PME */
6290 /* We do not require separate communicators */
6291 cr->mpi_comm_mygroup = cr->mpi_comm_mysim;
6295 if (cr->duty & DUTY_PP)
6297 /* Copy or make a new PP communicator */
6298 make_pp_communicator(fplog, cr, CartReorder);
6302 receive_ddindex2simnodeid(cr);
6305 if (!(cr->duty & DUTY_PME))
6307 /* Set up the commnuication to our PME node */
6308 dd->pme_nodeid = dd_simnode2pmenode(cr, cr->sim_nodeid);
6309 dd->pme_receive_vir_ener = receive_vir_ener(cr);
6312 fprintf(debug, "My pme_nodeid %d receive ener %d\n",
6313 dd->pme_nodeid, dd->pme_receive_vir_ener);
6318 dd->pme_nodeid = -1;
6323 dd->ma = init_gmx_domdec_master_t(dd,
6325 comm->cgs_gl.index[comm->cgs_gl.nr]);
6329 static real *get_slb_frac(FILE *fplog, const char *dir, int nc, const char *size_string)
6331 real *slb_frac, tot;
6336 if (nc > 1 && size_string != NULL)
6340 fprintf(fplog, "Using static load balancing for the %s direction\n",
6345 for (i = 0; i < nc; i++)
6348 sscanf(size_string, "%lf%n", &dbl, &n);
6351 gmx_fatal(FARGS, "Incorrect or not enough DD cell size entries for direction %s: '%s'", dir, size_string);
6360 fprintf(fplog, "Relative cell sizes:");
6362 for (i = 0; i < nc; i++)
6367 fprintf(fplog, " %5.3f", slb_frac[i]);
6372 fprintf(fplog, "\n");
6379 static int multi_body_bondeds_count(gmx_mtop_t *mtop)
6382 gmx_mtop_ilistloop_t iloop;
6386 iloop = gmx_mtop_ilistloop_init(mtop);
6387 while (gmx_mtop_ilistloop_next(iloop, &il, &nmol))
6389 for (ftype = 0; ftype < F_NRE; ftype++)
6391 if ((interaction_function[ftype].flags & IF_BOND) &&
6394 n += nmol*il[ftype].nr/(1 + NRAL(ftype));
6402 static int dd_getenv(FILE *fplog, const char *env_var, int def)
6408 val = getenv(env_var);
6411 if (sscanf(val, "%d", &nst) <= 0)
6417 fprintf(fplog, "Found env.var. %s = %s, using value %d\n",
6425 static void dd_warning(t_commrec *cr, FILE *fplog, const char *warn_string)
6429 fprintf(stderr, "\n%s\n", warn_string);
6433 fprintf(fplog, "\n%s\n", warn_string);
6437 static void check_dd_restrictions(t_commrec *cr, gmx_domdec_t *dd,
6438 t_inputrec *ir, FILE *fplog)
6440 if (ir->ePBC == epbcSCREW &&
6441 (dd->nc[XX] == 1 || dd->nc[YY] > 1 || dd->nc[ZZ] > 1))
6443 gmx_fatal(FARGS, "With pbc=%s can only do domain decomposition in the x-direction", epbc_names[ir->ePBC]);
6446 if (ir->ns_type == ensSIMPLE)
6448 gmx_fatal(FARGS, "Domain decomposition does not support simple neighbor searching, use grid searching or run with one MPI rank");
6451 if (ir->nstlist == 0)
6453 gmx_fatal(FARGS, "Domain decomposition does not work with nstlist=0");
6456 if (ir->comm_mode == ecmANGULAR && ir->ePBC != epbcNONE)
6458 dd_warning(cr, fplog, "comm-mode angular will give incorrect results when the comm group partially crosses a periodic boundary");
6462 static real average_cellsize_min(gmx_domdec_t *dd, gmx_ddbox_t *ddbox)
6467 r = ddbox->box_size[XX];
6468 for (di = 0; di < dd->ndim; di++)
6471 /* Check using the initial average cell size */
6472 r = min(r, ddbox->box_size[d]*ddbox->skew_fac[d]/dd->nc[d]);
6478 static int check_dlb_support(FILE *fplog, t_commrec *cr,
6479 const char *dlb_opt, gmx_bool bRecordLoad,
6480 unsigned long Flags, t_inputrec *ir)
6488 case 'a': eDLB = edlbAUTO; break;
6489 case 'n': eDLB = edlbNO; break;
6490 case 'y': eDLB = edlbYES; break;
6491 default: gmx_incons("Unknown dlb_opt");
6494 if (Flags & MD_RERUN)
6499 if (!EI_DYNAMICS(ir->eI))
6501 if (eDLB == edlbYES)
6503 sprintf(buf, "NOTE: dynamic load balancing is only supported with dynamics, not with integrator '%s'\n", EI(ir->eI));
6504 dd_warning(cr, fplog, buf);
6512 dd_warning(cr, fplog, "NOTE: Cycle counting is not supported on this architecture, will not use dynamic load balancing\n");
6517 if (Flags & MD_REPRODUCIBLE)
6524 dd_warning(cr, fplog, "NOTE: reproducibility requested, will not use dynamic load balancing\n");
6528 dd_warning(cr, fplog, "WARNING: reproducibility requested with dynamic load balancing, the simulation will NOT be binary reproducible\n");
6531 gmx_fatal(FARGS, "Death horror: undefined case (%d) for load balancing choice", eDLB);
6539 static void set_dd_dim(FILE *fplog, gmx_domdec_t *dd)
6544 if (getenv("GMX_DD_ORDER_ZYX") != NULL)
6546 /* Decomposition order z,y,x */
6549 fprintf(fplog, "Using domain decomposition order z, y, x\n");
6551 for (dim = DIM-1; dim >= 0; dim--)
6553 if (dd->nc[dim] > 1)
6555 dd->dim[dd->ndim++] = dim;
6561 /* Decomposition order x,y,z */
6562 for (dim = 0; dim < DIM; dim++)
6564 if (dd->nc[dim] > 1)
6566 dd->dim[dd->ndim++] = dim;
6572 static gmx_domdec_comm_t *init_dd_comm()
6574 gmx_domdec_comm_t *comm;
6578 snew(comm->cggl_flag, DIM*2);
6579 snew(comm->cgcm_state, DIM*2);
6580 for (i = 0; i < DIM*2; i++)
6582 comm->cggl_flag_nalloc[i] = 0;
6583 comm->cgcm_state_nalloc[i] = 0;
6586 comm->nalloc_int = 0;
6587 comm->buf_int = NULL;
6589 vec_rvec_init(&comm->vbuf);
6591 comm->n_load_have = 0;
6592 comm->n_load_collect = 0;
6594 for (i = 0; i < ddnatNR-ddnatZONE; i++)
6596 comm->sum_nat[i] = 0;
6600 comm->load_step = 0;
6603 clear_ivec(comm->load_lim);
6610 gmx_domdec_t *init_domain_decomposition(FILE *fplog, t_commrec *cr,
6611 unsigned long Flags,
6613 real comm_distance_min, real rconstr,
6614 const char *dlb_opt, real dlb_scale,
6615 const char *sizex, const char *sizey, const char *sizez,
6616 gmx_mtop_t *mtop, t_inputrec *ir,
6617 matrix box, rvec *x,
6619 int *npme_x, int *npme_y)
6622 gmx_domdec_comm_t *comm;
6625 real r_2b, r_mb, r_bonded = -1, r_bonded_limit = -1, limit, acs;
6632 "\nInitializing Domain Decomposition on %d ranks\n", cr->nnodes);
6637 dd->comm = init_dd_comm();
6639 snew(comm->cggl_flag, DIM*2);
6640 snew(comm->cgcm_state, DIM*2);
6642 dd->npbcdim = ePBC2npbcdim(ir->ePBC);
6643 dd->bScrewPBC = (ir->ePBC == epbcSCREW);
6645 dd->bSendRecv2 = dd_getenv(fplog, "GMX_DD_USE_SENDRECV2", 0);
6646 comm->dlb_scale_lim = dd_getenv(fplog, "GMX_DLB_MAX_BOX_SCALING", 10);
6647 comm->eFlop = dd_getenv(fplog, "GMX_DLB_BASED_ON_FLOPS", 0);
6648 recload = dd_getenv(fplog, "GMX_DD_RECORD_LOAD", 1);
6649 comm->nstSortCG = dd_getenv(fplog, "GMX_DD_NST_SORT_CHARGE_GROUPS", 1);
6650 comm->nstDDDump = dd_getenv(fplog, "GMX_DD_NST_DUMP", 0);
6651 comm->nstDDDumpGrid = dd_getenv(fplog, "GMX_DD_NST_DUMP_GRID", 0);
6652 comm->DD_debug = dd_getenv(fplog, "GMX_DD_DEBUG", 0);
6654 dd->pme_recv_f_alloc = 0;
6655 dd->pme_recv_f_buf = NULL;
6657 if (dd->bSendRecv2 && fplog)
6659 fprintf(fplog, "Will use two sequential MPI_Sendrecv calls instead of two simultaneous non-blocking MPI_Irecv and MPI_Isend pairs for constraint and vsite communication\n");
6665 fprintf(fplog, "Will load balance based on FLOP count\n");
6667 if (comm->eFlop > 1)
6669 srand(1+cr->nodeid);
6671 comm->bRecordLoad = TRUE;
6675 comm->bRecordLoad = (wallcycle_have_counter() && recload > 0);
6679 /* Initialize to GPU share count to 0, might change later */
6680 comm->nrank_gpu_shared = 0;
6682 comm->eDLB = check_dlb_support(fplog, cr, dlb_opt, comm->bRecordLoad, Flags, ir);
6684 comm->bDynLoadBal = (comm->eDLB == edlbYES);
6687 fprintf(fplog, "Dynamic load balancing: %s\n", edlb_names[comm->eDLB]);
6689 dd->bGridJump = comm->bDynLoadBal;
6690 comm->bPMELoadBalDLBLimits = FALSE;
6692 if (comm->nstSortCG)
6696 if (comm->nstSortCG == 1)
6698 fprintf(fplog, "Will sort the charge groups at every domain (re)decomposition\n");
6702 fprintf(fplog, "Will sort the charge groups every %d steps\n",
6706 snew(comm->sort, 1);
6712 fprintf(fplog, "Will not sort the charge groups\n");
6716 comm->bCGs = (ncg_mtop(mtop) < mtop->natoms);
6718 comm->bInterCGBondeds = (ncg_mtop(mtop) > mtop->mols.nr);
6719 if (comm->bInterCGBondeds)
6721 comm->bInterCGMultiBody = (multi_body_bondeds_count(mtop) > 0);
6725 comm->bInterCGMultiBody = FALSE;
6728 dd->bInterCGcons = inter_charge_group_constraints(mtop);
6729 dd->bInterCGsettles = inter_charge_group_settles(mtop);
6731 if (ir->rlistlong == 0)
6733 /* Set the cut-off to some very large value,
6734 * so we don't need if statements everywhere in the code.
6735 * We use sqrt, since the cut-off is squared in some places.
6737 comm->cutoff = GMX_CUTOFF_INF;
6741 comm->cutoff = ir->rlistlong;
6743 comm->cutoff_mbody = 0;
6745 comm->cellsize_limit = 0;
6746 comm->bBondComm = FALSE;
6748 if (comm->bInterCGBondeds)
6750 if (comm_distance_min > 0)
6752 comm->cutoff_mbody = comm_distance_min;
6753 if (Flags & MD_DDBONDCOMM)
6755 comm->bBondComm = (comm->cutoff_mbody > comm->cutoff);
6759 comm->cutoff = max(comm->cutoff, comm->cutoff_mbody);
6761 r_bonded_limit = comm->cutoff_mbody;
6763 else if (ir->bPeriodicMols)
6765 /* Can not easily determine the required cut-off */
6766 dd_warning(cr, fplog, "NOTE: Periodic molecules are present in this system. Because of this, the domain decomposition algorithm cannot easily determine the minimum cell size that it requires for treating bonded interactions. Instead, domain decomposition will assume that half the non-bonded cut-off will be a suitable lower bound.\n");
6767 comm->cutoff_mbody = comm->cutoff/2;
6768 r_bonded_limit = comm->cutoff_mbody;
6774 dd_bonded_cg_distance(fplog, mtop, ir, x, box,
6775 Flags & MD_DDBONDCHECK, &r_2b, &r_mb);
6777 gmx_bcast(sizeof(r_2b), &r_2b, cr);
6778 gmx_bcast(sizeof(r_mb), &r_mb, cr);
6780 /* We use an initial margin of 10% for the minimum cell size,
6781 * except when we are just below the non-bonded cut-off.
6783 if (Flags & MD_DDBONDCOMM)
6785 if (max(r_2b, r_mb) > comm->cutoff)
6787 r_bonded = max(r_2b, r_mb);
6788 r_bonded_limit = 1.1*r_bonded;
6789 comm->bBondComm = TRUE;
6794 r_bonded_limit = min(1.1*r_bonded, comm->cutoff);
6796 /* We determine cutoff_mbody later */
6800 /* No special bonded communication,
6801 * simply increase the DD cut-off.
6803 r_bonded_limit = 1.1*max(r_2b, r_mb);
6804 comm->cutoff_mbody = r_bonded_limit;
6805 comm->cutoff = max(comm->cutoff, comm->cutoff_mbody);
6808 comm->cellsize_limit = max(comm->cellsize_limit, r_bonded_limit);
6812 "Minimum cell size due to bonded interactions: %.3f nm\n",
6813 comm->cellsize_limit);
6817 if (dd->bInterCGcons && rconstr <= 0)
6819 /* There is a cell size limit due to the constraints (P-LINCS) */
6820 rconstr = constr_r_max(fplog, mtop, ir);
6824 "Estimated maximum distance required for P-LINCS: %.3f nm\n",
6826 if (rconstr > comm->cellsize_limit)
6828 fprintf(fplog, "This distance will limit the DD cell size, you can override this with -rcon\n");
6832 else if (rconstr > 0 && fplog)
6834 /* Here we do not check for dd->bInterCGcons,
6835 * because one can also set a cell size limit for virtual sites only
6836 * and at this point we don't know yet if there are intercg v-sites.
6839 "User supplied maximum distance required for P-LINCS: %.3f nm\n",
6842 comm->cellsize_limit = max(comm->cellsize_limit, rconstr);
6844 comm->cgs_gl = gmx_mtop_global_cgs(mtop);
6848 copy_ivec(nc, dd->nc);
6849 set_dd_dim(fplog, dd);
6850 set_ddbox_cr(cr, &dd->nc, ir, box, &comm->cgs_gl, x, ddbox);
6852 if (cr->npmenodes == -1)
6856 acs = average_cellsize_min(dd, ddbox);
6857 if (acs < comm->cellsize_limit)
6861 fprintf(fplog, "ERROR: The initial cell size (%f) is smaller than the cell size limit (%f)\n", acs, comm->cellsize_limit);
6863 gmx_fatal_collective(FARGS, cr, NULL,
6864 "The initial cell size (%f) is smaller than the cell size limit (%f), change options -dd, -rdd or -rcon, see the log file for details",
6865 acs, comm->cellsize_limit);
6870 set_ddbox_cr(cr, NULL, ir, box, &comm->cgs_gl, x, ddbox);
6872 /* We need to choose the optimal DD grid and possibly PME nodes */
6873 limit = dd_choose_grid(fplog, cr, dd, ir, mtop, box, ddbox,
6874 comm->eDLB != edlbNO, dlb_scale,
6875 comm->cellsize_limit, comm->cutoff,
6876 comm->bInterCGBondeds);
6878 if (dd->nc[XX] == 0)
6880 bC = (dd->bInterCGcons && rconstr > r_bonded_limit);
6881 sprintf(buf, "Change the number of ranks or mdrun option %s%s%s",
6882 !bC ? "-rdd" : "-rcon",
6883 comm->eDLB != edlbNO ? " or -dds" : "",
6884 bC ? " or your LINCS settings" : "");
6886 gmx_fatal_collective(FARGS, cr, NULL,
6887 "There is no domain decomposition for %d ranks that is compatible with the given box and a minimum cell size of %g nm\n"
6889 "Look in the log file for details on the domain decomposition",
6890 cr->nnodes-cr->npmenodes, limit, buf);
6892 set_dd_dim(fplog, dd);
6898 "Domain decomposition grid %d x %d x %d, separate PME ranks %d\n",
6899 dd->nc[XX], dd->nc[YY], dd->nc[ZZ], cr->npmenodes);
6902 dd->nnodes = dd->nc[XX]*dd->nc[YY]*dd->nc[ZZ];
6903 if (cr->nnodes - dd->nnodes != cr->npmenodes)
6905 gmx_fatal_collective(FARGS, cr, NULL,
6906 "The size of the domain decomposition grid (%d) does not match the number of ranks (%d). The total number of ranks is %d",
6907 dd->nnodes, cr->nnodes - cr->npmenodes, cr->nnodes);
6909 if (cr->npmenodes > dd->nnodes)
6911 gmx_fatal_collective(FARGS, cr, NULL,
6912 "The number of separate PME ranks (%d) is larger than the number of PP ranks (%d), this is not supported.", cr->npmenodes, dd->nnodes);
6914 if (cr->npmenodes > 0)
6916 comm->npmenodes = cr->npmenodes;
6920 comm->npmenodes = dd->nnodes;
6923 if (EEL_PME(ir->coulombtype) || EVDW_PME(ir->vdwtype))
6925 /* The following choices should match those
6926 * in comm_cost_est in domdec_setup.c.
6927 * Note that here the checks have to take into account
6928 * that the decomposition might occur in a different order than xyz
6929 * (for instance through the env.var. GMX_DD_ORDER_ZYX),
6930 * in which case they will not match those in comm_cost_est,
6931 * but since that is mainly for testing purposes that's fine.
6933 if (dd->ndim >= 2 && dd->dim[0] == XX && dd->dim[1] == YY &&
6934 comm->npmenodes > dd->nc[XX] && comm->npmenodes % dd->nc[XX] == 0 &&
6935 getenv("GMX_PMEONEDD") == NULL)
6937 comm->npmedecompdim = 2;
6938 comm->npmenodes_x = dd->nc[XX];
6939 comm->npmenodes_y = comm->npmenodes/comm->npmenodes_x;
6943 /* In case nc is 1 in both x and y we could still choose to
6944 * decompose pme in y instead of x, but we use x for simplicity.
6946 comm->npmedecompdim = 1;
6947 if (dd->dim[0] == YY)
6949 comm->npmenodes_x = 1;
6950 comm->npmenodes_y = comm->npmenodes;
6954 comm->npmenodes_x = comm->npmenodes;
6955 comm->npmenodes_y = 1;
6960 fprintf(fplog, "PME domain decomposition: %d x %d x %d\n",
6961 comm->npmenodes_x, comm->npmenodes_y, 1);
6966 comm->npmedecompdim = 0;
6967 comm->npmenodes_x = 0;
6968 comm->npmenodes_y = 0;
6971 /* Technically we don't need both of these,
6972 * but it simplifies code not having to recalculate it.
6974 *npme_x = comm->npmenodes_x;
6975 *npme_y = comm->npmenodes_y;
6977 snew(comm->slb_frac, DIM);
6978 if (comm->eDLB == edlbNO)
6980 comm->slb_frac[XX] = get_slb_frac(fplog, "x", dd->nc[XX], sizex);
6981 comm->slb_frac[YY] = get_slb_frac(fplog, "y", dd->nc[YY], sizey);
6982 comm->slb_frac[ZZ] = get_slb_frac(fplog, "z", dd->nc[ZZ], sizez);
6985 if (comm->bInterCGBondeds && comm->cutoff_mbody == 0)
6987 if (comm->bBondComm || comm->eDLB != edlbNO)
6989 /* Set the bonded communication distance to halfway
6990 * the minimum and the maximum,
6991 * since the extra communication cost is nearly zero.
6993 acs = average_cellsize_min(dd, ddbox);
6994 comm->cutoff_mbody = 0.5*(r_bonded + acs);
6995 if (comm->eDLB != edlbNO)
6997 /* Check if this does not limit the scaling */
6998 comm->cutoff_mbody = min(comm->cutoff_mbody, dlb_scale*acs);
7000 if (!comm->bBondComm)
7002 /* Without bBondComm do not go beyond the n.b. cut-off */
7003 comm->cutoff_mbody = min(comm->cutoff_mbody, comm->cutoff);
7004 if (comm->cellsize_limit >= comm->cutoff)
7006 /* We don't loose a lot of efficieny
7007 * when increasing it to the n.b. cut-off.
7008 * It can even be slightly faster, because we need
7009 * less checks for the communication setup.
7011 comm->cutoff_mbody = comm->cutoff;
7014 /* Check if we did not end up below our original limit */
7015 comm->cutoff_mbody = max(comm->cutoff_mbody, r_bonded_limit);
7017 if (comm->cutoff_mbody > comm->cellsize_limit)
7019 comm->cellsize_limit = comm->cutoff_mbody;
7022 /* Without DLB and cutoff_mbody<cutoff, cutoff_mbody is dynamic */
7027 fprintf(debug, "Bonded atom communication beyond the cut-off: %d\n"
7028 "cellsize limit %f\n",
7029 comm->bBondComm, comm->cellsize_limit);
7034 check_dd_restrictions(cr, dd, ir, fplog);
7037 comm->partition_step = INT_MIN;
7040 clear_dd_cycle_counts(dd);
7045 static void set_dlb_limits(gmx_domdec_t *dd)
7050 for (d = 0; d < dd->ndim; d++)
7052 dd->comm->cd[d].np = dd->comm->cd[d].np_dlb;
7053 dd->comm->cellsize_min[dd->dim[d]] =
7054 dd->comm->cellsize_min_dlb[dd->dim[d]];
7059 static void turn_on_dlb(FILE *fplog, t_commrec *cr, gmx_int64_t step)
7062 gmx_domdec_comm_t *comm;
7072 fprintf(fplog, "At step %s the performance loss due to force load imbalance is %.1f %%\n", gmx_step_str(step, buf), dd_force_imb_perf_loss(dd)*100);
7075 cellsize_min = comm->cellsize_min[dd->dim[0]];
7076 for (d = 1; d < dd->ndim; d++)
7078 cellsize_min = min(cellsize_min, comm->cellsize_min[dd->dim[d]]);
7081 if (cellsize_min < comm->cellsize_limit*1.05)
7083 dd_warning(cr, fplog, "NOTE: the minimum cell size is smaller than 1.05 times the cell size limit, will not turn on dynamic load balancing\n");
7085 /* Change DLB from "auto" to "no". */
7086 comm->eDLB = edlbNO;
7091 dd_warning(cr, fplog, "NOTE: Turning on dynamic load balancing\n");
7092 comm->bDynLoadBal = TRUE;
7093 dd->bGridJump = TRUE;
7097 /* We can set the required cell size info here,
7098 * so we do not need to communicate this.
7099 * The grid is completely uniform.
7101 for (d = 0; d < dd->ndim; d++)
7105 comm->load[d].sum_m = comm->load[d].sum;
7107 nc = dd->nc[dd->dim[d]];
7108 for (i = 0; i < nc; i++)
7110 comm->root[d]->cell_f[i] = i/(real)nc;
7113 comm->root[d]->cell_f_max0[i] = i /(real)nc;
7114 comm->root[d]->cell_f_min1[i] = (i+1)/(real)nc;
7117 comm->root[d]->cell_f[nc] = 1.0;
7122 static char *init_bLocalCG(gmx_mtop_t *mtop)
7127 ncg = ncg_mtop(mtop);
7128 snew(bLocalCG, ncg);
7129 for (cg = 0; cg < ncg; cg++)
7131 bLocalCG[cg] = FALSE;
7137 void dd_init_bondeds(FILE *fplog,
7138 gmx_domdec_t *dd, gmx_mtop_t *mtop,
7140 t_inputrec *ir, gmx_bool bBCheck, cginfo_mb_t *cginfo_mb)
7142 gmx_domdec_comm_t *comm;
7146 dd_make_reverse_top(fplog, dd, mtop, vsite, ir, bBCheck);
7150 if (comm->bBondComm)
7152 /* Communicate atoms beyond the cut-off for bonded interactions */
7155 comm->cglink = make_charge_group_links(mtop, dd, cginfo_mb);
7157 comm->bLocalCG = init_bLocalCG(mtop);
7161 /* Only communicate atoms based on cut-off */
7162 comm->cglink = NULL;
7163 comm->bLocalCG = NULL;
7167 static void print_dd_settings(FILE *fplog, gmx_domdec_t *dd,
7169 gmx_bool bDynLoadBal, real dlb_scale,
7172 gmx_domdec_comm_t *comm;
7187 fprintf(fplog, "The maximum number of communication pulses is:");
7188 for (d = 0; d < dd->ndim; d++)
7190 fprintf(fplog, " %c %d", dim2char(dd->dim[d]), comm->cd[d].np_dlb);
7192 fprintf(fplog, "\n");
7193 fprintf(fplog, "The minimum size for domain decomposition cells is %.3f nm\n", comm->cellsize_limit);
7194 fprintf(fplog, "The requested allowed shrink of DD cells (option -dds) is: %.2f\n", dlb_scale);
7195 fprintf(fplog, "The allowed shrink of domain decomposition cells is:");
7196 for (d = 0; d < DIM; d++)
7200 if (d >= ddbox->npbcdim && dd->nc[d] == 2)
7207 comm->cellsize_min_dlb[d]/
7208 (ddbox->box_size[d]*ddbox->skew_fac[d]/dd->nc[d]);
7210 fprintf(fplog, " %c %.2f", dim2char(d), shrink);
7213 fprintf(fplog, "\n");
7217 set_dd_cell_sizes_slb(dd, ddbox, setcellsizeslbPULSE_ONLY, np);
7218 fprintf(fplog, "The initial number of communication pulses is:");
7219 for (d = 0; d < dd->ndim; d++)
7221 fprintf(fplog, " %c %d", dim2char(dd->dim[d]), np[dd->dim[d]]);
7223 fprintf(fplog, "\n");
7224 fprintf(fplog, "The initial domain decomposition cell size is:");
7225 for (d = 0; d < DIM; d++)
7229 fprintf(fplog, " %c %.2f nm",
7230 dim2char(d), dd->comm->cellsize_min[d]);
7233 fprintf(fplog, "\n\n");
7236 if (comm->bInterCGBondeds || dd->vsite_comm || dd->constraint_comm)
7238 fprintf(fplog, "The maximum allowed distance for charge groups involved in interactions is:\n");
7239 fprintf(fplog, "%40s %-7s %6.3f nm\n",
7240 "non-bonded interactions", "", comm->cutoff);
7244 limit = dd->comm->cellsize_limit;
7248 if (dynamic_dd_box(ddbox, ir))
7250 fprintf(fplog, "(the following are initial values, they could change due to box deformation)\n");
7252 limit = dd->comm->cellsize_min[XX];
7253 for (d = 1; d < DIM; d++)
7255 limit = min(limit, dd->comm->cellsize_min[d]);
7259 if (comm->bInterCGBondeds)
7261 fprintf(fplog, "%40s %-7s %6.3f nm\n",
7262 "two-body bonded interactions", "(-rdd)",
7263 max(comm->cutoff, comm->cutoff_mbody));
7264 fprintf(fplog, "%40s %-7s %6.3f nm\n",
7265 "multi-body bonded interactions", "(-rdd)",
7266 (comm->bBondComm || dd->bGridJump) ? comm->cutoff_mbody : min(comm->cutoff, limit));
7270 fprintf(fplog, "%40s %-7s %6.3f nm\n",
7271 "virtual site constructions", "(-rcon)", limit);
7273 if (dd->constraint_comm)
7275 sprintf(buf, "atoms separated by up to %d constraints",
7277 fprintf(fplog, "%40s %-7s %6.3f nm\n",
7278 buf, "(-rcon)", limit);
7280 fprintf(fplog, "\n");
7286 static void set_cell_limits_dlb(gmx_domdec_t *dd,
7288 const t_inputrec *ir,
7289 const gmx_ddbox_t *ddbox)
7291 gmx_domdec_comm_t *comm;
7292 int d, dim, npulse, npulse_d_max, npulse_d;
7297 bNoCutOff = (ir->rvdw == 0 || ir->rcoulomb == 0);
7299 /* Determine the maximum number of comm. pulses in one dimension */
7301 comm->cellsize_limit = max(comm->cellsize_limit, comm->cutoff_mbody);
7303 /* Determine the maximum required number of grid pulses */
7304 if (comm->cellsize_limit >= comm->cutoff)
7306 /* Only a single pulse is required */
7309 else if (!bNoCutOff && comm->cellsize_limit > 0)
7311 /* We round down slightly here to avoid overhead due to the latency
7312 * of extra communication calls when the cut-off
7313 * would be only slightly longer than the cell size.
7314 * Later cellsize_limit is redetermined,
7315 * so we can not miss interactions due to this rounding.
7317 npulse = (int)(0.96 + comm->cutoff/comm->cellsize_limit);
7321 /* There is no cell size limit */
7322 npulse = max(dd->nc[XX]-1, max(dd->nc[YY]-1, dd->nc[ZZ]-1));
7325 if (!bNoCutOff && npulse > 1)
7327 /* See if we can do with less pulses, based on dlb_scale */
7329 for (d = 0; d < dd->ndim; d++)
7332 npulse_d = (int)(1 + dd->nc[dim]*comm->cutoff
7333 /(ddbox->box_size[dim]*ddbox->skew_fac[dim]*dlb_scale));
7334 npulse_d_max = max(npulse_d_max, npulse_d);
7336 npulse = min(npulse, npulse_d_max);
7339 /* This env var can override npulse */
7340 d = dd_getenv(debug, "GMX_DD_NPULSE", 0);
7347 comm->bVacDLBNoLimit = (ir->ePBC == epbcNONE);
7348 for (d = 0; d < dd->ndim; d++)
7350 comm->cd[d].np_dlb = min(npulse, dd->nc[dd->dim[d]]-1);
7351 comm->cd[d].np_nalloc = comm->cd[d].np_dlb;
7352 snew(comm->cd[d].ind, comm->cd[d].np_nalloc);
7353 comm->maxpulse = max(comm->maxpulse, comm->cd[d].np_dlb);
7354 if (comm->cd[d].np_dlb < dd->nc[dd->dim[d]]-1)
7356 comm->bVacDLBNoLimit = FALSE;
7360 /* cellsize_limit is set for LINCS in init_domain_decomposition */
7361 if (!comm->bVacDLBNoLimit)
7363 comm->cellsize_limit = max(comm->cellsize_limit,
7364 comm->cutoff/comm->maxpulse);
7366 comm->cellsize_limit = max(comm->cellsize_limit, comm->cutoff_mbody);
7367 /* Set the minimum cell size for each DD dimension */
7368 for (d = 0; d < dd->ndim; d++)
7370 if (comm->bVacDLBNoLimit ||
7371 comm->cd[d].np_dlb*comm->cellsize_limit >= comm->cutoff)
7373 comm->cellsize_min_dlb[dd->dim[d]] = comm->cellsize_limit;
7377 comm->cellsize_min_dlb[dd->dim[d]] =
7378 comm->cutoff/comm->cd[d].np_dlb;
7381 if (comm->cutoff_mbody <= 0)
7383 comm->cutoff_mbody = min(comm->cutoff, comm->cellsize_limit);
7385 if (comm->bDynLoadBal)
7391 gmx_bool dd_bonded_molpbc(gmx_domdec_t *dd, int ePBC)
7393 /* If each molecule is a single charge group
7394 * or we use domain decomposition for each periodic dimension,
7395 * we do not need to take pbc into account for the bonded interactions.
7397 return (ePBC != epbcNONE && dd->comm->bInterCGBondeds &&
7400 (dd->nc[ZZ] > 1 || ePBC == epbcXY)));
7403 void set_dd_parameters(FILE *fplog, gmx_domdec_t *dd, real dlb_scale,
7404 t_inputrec *ir, gmx_ddbox_t *ddbox)
7406 gmx_domdec_comm_t *comm;
7412 /* Initialize the thread data.
7413 * This can not be done in init_domain_decomposition,
7414 * as the numbers of threads is determined later.
7416 comm->nth = gmx_omp_nthreads_get(emntDomdec);
7419 snew(comm->dth, comm->nth);
7422 if (EEL_PME(ir->coulombtype) || EVDW_PME(ir->vdwtype))
7424 init_ddpme(dd, &comm->ddpme[0], 0);
7425 if (comm->npmedecompdim >= 2)
7427 init_ddpme(dd, &comm->ddpme[1], 1);
7432 comm->npmenodes = 0;
7433 if (dd->pme_nodeid >= 0)
7435 gmx_fatal_collective(FARGS, NULL, dd,
7436 "Can not have separate PME ranks without PME electrostatics");
7442 fprintf(debug, "The DD cut-off is %f\n", comm->cutoff);
7444 if (comm->eDLB != edlbNO)
7446 set_cell_limits_dlb(dd, dlb_scale, ir, ddbox);
7449 print_dd_settings(fplog, dd, ir, comm->bDynLoadBal, dlb_scale, ddbox);
7450 if (comm->eDLB == edlbAUTO)
7454 fprintf(fplog, "When dynamic load balancing gets turned on, these settings will change to:\n");
7456 print_dd_settings(fplog, dd, ir, TRUE, dlb_scale, ddbox);
7459 if (ir->ePBC == epbcNONE)
7461 vol_frac = 1 - 1/(double)dd->nnodes;
7466 (1 + comm_box_frac(dd->nc, comm->cutoff, ddbox))/(double)dd->nnodes;
7470 fprintf(debug, "Volume fraction for all DD zones: %f\n", vol_frac);
7472 natoms_tot = comm->cgs_gl.index[comm->cgs_gl.nr];
7474 dd->ga2la = ga2la_init(natoms_tot, vol_frac*natoms_tot);
7477 static gmx_bool test_dd_cutoff(t_commrec *cr,
7478 t_state *state, t_inputrec *ir,
7489 set_ddbox(dd, FALSE, cr, ir, state->box,
7490 TRUE, &dd->comm->cgs_gl, state->x, &ddbox);
7494 for (d = 0; d < dd->ndim; d++)
7498 inv_cell_size = DD_CELL_MARGIN*dd->nc[dim]/ddbox.box_size[dim];
7499 if (dynamic_dd_box(&ddbox, ir))
7501 inv_cell_size *= DD_PRES_SCALE_MARGIN;
7504 np = 1 + (int)(cutoff_req*inv_cell_size*ddbox.skew_fac[dim]);
7506 if (dd->comm->eDLB != edlbNO && dim < ddbox.npbcdim &&
7507 dd->comm->cd[d].np_dlb > 0)
7509 if (np > dd->comm->cd[d].np_dlb)
7514 /* If a current local cell size is smaller than the requested
7515 * cut-off, we could still fix it, but this gets very complicated.
7516 * Without fixing here, we might actually need more checks.
7518 if ((dd->comm->cell_x1[dim] - dd->comm->cell_x0[dim])*ddbox.skew_fac[dim]*dd->comm->cd[d].np_dlb < cutoff_req)
7525 if (dd->comm->eDLB != edlbNO)
7527 /* If DLB is not active yet, we don't need to check the grid jumps.
7528 * Actually we shouldn't, because then the grid jump data is not set.
7530 if (dd->comm->bDynLoadBal &&
7531 check_grid_jump(0, dd, cutoff_req, &ddbox, FALSE))
7536 gmx_sumi(1, &LocallyLimited, cr);
7538 if (LocallyLimited > 0)
7547 gmx_bool change_dd_cutoff(t_commrec *cr, t_state *state, t_inputrec *ir,
7550 gmx_bool bCutoffAllowed;
7552 bCutoffAllowed = test_dd_cutoff(cr, state, ir, cutoff_req);
7556 cr->dd->comm->cutoff = cutoff_req;
7559 return bCutoffAllowed;
7562 void change_dd_dlb_cutoff_limit(t_commrec *cr)
7564 gmx_domdec_comm_t *comm;
7566 comm = cr->dd->comm;
7568 /* Turn on the DLB limiting (might have been on already) */
7569 comm->bPMELoadBalDLBLimits = TRUE;
7571 /* Change the cut-off limit */
7572 comm->PMELoadBal_max_cutoff = comm->cutoff;
7575 static void merge_cg_buffers(int ncell,
7576 gmx_domdec_comm_dim_t *cd, int pulse,
7578 int *index_gl, int *recv_i,
7579 rvec *cg_cm, rvec *recv_vr,
7581 cginfo_mb_t *cginfo_mb, int *cginfo)
7583 gmx_domdec_ind_t *ind, *ind_p;
7584 int p, cell, c, cg, cg0, cg1, cg_gl, nat;
7585 int shift, shift_at;
7587 ind = &cd->ind[pulse];
7589 /* First correct the already stored data */
7590 shift = ind->nrecv[ncell];
7591 for (cell = ncell-1; cell >= 0; cell--)
7593 shift -= ind->nrecv[cell];
7596 /* Move the cg's present from previous grid pulses */
7597 cg0 = ncg_cell[ncell+cell];
7598 cg1 = ncg_cell[ncell+cell+1];
7599 cgindex[cg1+shift] = cgindex[cg1];
7600 for (cg = cg1-1; cg >= cg0; cg--)
7602 index_gl[cg+shift] = index_gl[cg];
7603 copy_rvec(cg_cm[cg], cg_cm[cg+shift]);
7604 cgindex[cg+shift] = cgindex[cg];
7605 cginfo[cg+shift] = cginfo[cg];
7607 /* Correct the already stored send indices for the shift */
7608 for (p = 1; p <= pulse; p++)
7610 ind_p = &cd->ind[p];
7612 for (c = 0; c < cell; c++)
7614 cg0 += ind_p->nsend[c];
7616 cg1 = cg0 + ind_p->nsend[cell];
7617 for (cg = cg0; cg < cg1; cg++)
7619 ind_p->index[cg] += shift;
7625 /* Merge in the communicated buffers */
7629 for (cell = 0; cell < ncell; cell++)
7631 cg1 = ncg_cell[ncell+cell+1] + shift;
7634 /* Correct the old cg indices */
7635 for (cg = ncg_cell[ncell+cell]; cg < cg1; cg++)
7637 cgindex[cg+1] += shift_at;
7640 for (cg = 0; cg < ind->nrecv[cell]; cg++)
7642 /* Copy this charge group from the buffer */
7643 index_gl[cg1] = recv_i[cg0];
7644 copy_rvec(recv_vr[cg0], cg_cm[cg1]);
7645 /* Add it to the cgindex */
7646 cg_gl = index_gl[cg1];
7647 cginfo[cg1] = ddcginfo(cginfo_mb, cg_gl);
7648 nat = GET_CGINFO_NATOMS(cginfo[cg1]);
7649 cgindex[cg1+1] = cgindex[cg1] + nat;
7654 shift += ind->nrecv[cell];
7655 ncg_cell[ncell+cell+1] = cg1;
7659 static void make_cell2at_index(gmx_domdec_comm_dim_t *cd,
7660 int nzone, int cg0, const int *cgindex)
7664 /* Store the atom block boundaries for easy copying of communication buffers
7667 for (zone = 0; zone < nzone; zone++)
7669 for (p = 0; p < cd->np; p++)
7671 cd->ind[p].cell2at0[zone] = cgindex[cg];
7672 cg += cd->ind[p].nrecv[zone];
7673 cd->ind[p].cell2at1[zone] = cgindex[cg];
7678 static gmx_bool missing_link(t_blocka *link, int cg_gl, char *bLocalCG)
7684 for (i = link->index[cg_gl]; i < link->index[cg_gl+1]; i++)
7686 if (!bLocalCG[link->a[i]])
7695 /* Domain corners for communication, a maximum of 4 i-zones see a j domain */
7697 real c[DIM][4]; /* the corners for the non-bonded communication */
7698 real cr0; /* corner for rounding */
7699 real cr1[4]; /* corners for rounding */
7700 real bc[DIM]; /* corners for bounded communication */
7701 real bcr1; /* corner for rounding for bonded communication */
7704 /* Determine the corners of the domain(s) we are communicating with */
7706 set_dd_corners(const gmx_domdec_t *dd,
7707 int dim0, int dim1, int dim2,
7711 const gmx_domdec_comm_t *comm;
7712 const gmx_domdec_zones_t *zones;
7717 zones = &comm->zones;
7719 /* Keep the compiler happy */
7723 /* The first dimension is equal for all cells */
7724 c->c[0][0] = comm->cell_x0[dim0];
7727 c->bc[0] = c->c[0][0];
7732 /* This cell row is only seen from the first row */
7733 c->c[1][0] = comm->cell_x0[dim1];
7734 /* All rows can see this row */
7735 c->c[1][1] = comm->cell_x0[dim1];
7738 c->c[1][1] = max(comm->cell_x0[dim1], comm->zone_d1[1].mch0);
7741 /* For the multi-body distance we need the maximum */
7742 c->bc[1] = max(comm->cell_x0[dim1], comm->zone_d1[1].p1_0);
7745 /* Set the upper-right corner for rounding */
7746 c->cr0 = comm->cell_x1[dim0];
7751 for (j = 0; j < 4; j++)
7753 c->c[2][j] = comm->cell_x0[dim2];
7757 /* Use the maximum of the i-cells that see a j-cell */
7758 for (i = 0; i < zones->nizone; i++)
7760 for (j = zones->izone[i].j0; j < zones->izone[i].j1; j++)
7766 comm->zone_d2[zones->shift[i][dim0]][zones->shift[i][dim1]].mch0);
7772 /* For the multi-body distance we need the maximum */
7773 c->bc[2] = comm->cell_x0[dim2];
7774 for (i = 0; i < 2; i++)
7776 for (j = 0; j < 2; j++)
7778 c->bc[2] = max(c->bc[2], comm->zone_d2[i][j].p1_0);
7784 /* Set the upper-right corner for rounding */
7785 /* Cell (0,0,0) and cell (1,0,0) can see cell 4 (0,1,1)
7786 * Only cell (0,0,0) can see cell 7 (1,1,1)
7788 c->cr1[0] = comm->cell_x1[dim1];
7789 c->cr1[3] = comm->cell_x1[dim1];
7792 c->cr1[0] = max(comm->cell_x1[dim1], comm->zone_d1[1].mch1);
7795 /* For the multi-body distance we need the maximum */
7796 c->bcr1 = max(comm->cell_x1[dim1], comm->zone_d1[1].p1_1);
7803 /* Determine which cg's we need to send in this pulse from this zone */
7805 get_zone_pulse_cgs(gmx_domdec_t *dd,
7806 int zonei, int zone,
7808 const int *index_gl,
7810 int dim, int dim_ind,
7811 int dim0, int dim1, int dim2,
7812 real r_comm2, real r_bcomm2,
7816 real skew_fac2_d, real skew_fac_01,
7817 rvec *v_d, rvec *v_0, rvec *v_1,
7818 const dd_corners_t *c,
7820 gmx_bool bDistBonded,
7826 gmx_domdec_ind_t *ind,
7827 int **ibuf, int *ibuf_nalloc,
7833 gmx_domdec_comm_t *comm;
7835 gmx_bool bDistMB_pulse;
7837 real r2, rb2, r, tric_sh;
7840 int nsend_z, nsend, nat;
7844 bScrew = (dd->bScrewPBC && dim == XX);
7846 bDistMB_pulse = (bDistMB && bDistBonded);
7852 for (cg = cg0; cg < cg1; cg++)
7856 if (tric_dist[dim_ind] == 0)
7858 /* Rectangular direction, easy */
7859 r = cg_cm[cg][dim] - c->c[dim_ind][zone];
7866 r = cg_cm[cg][dim] - c->bc[dim_ind];
7872 /* Rounding gives at most a 16% reduction
7873 * in communicated atoms
7875 if (dim_ind >= 1 && (zonei == 1 || zonei == 2))
7877 r = cg_cm[cg][dim0] - c->cr0;
7878 /* This is the first dimension, so always r >= 0 */
7885 if (dim_ind == 2 && (zonei == 2 || zonei == 3))
7887 r = cg_cm[cg][dim1] - c->cr1[zone];
7894 r = cg_cm[cg][dim1] - c->bcr1;
7904 /* Triclinic direction, more complicated */
7907 /* Rounding, conservative as the skew_fac multiplication
7908 * will slightly underestimate the distance.
7910 if (dim_ind >= 1 && (zonei == 1 || zonei == 2))
7912 rn[dim0] = cg_cm[cg][dim0] - c->cr0;
7913 for (i = dim0+1; i < DIM; i++)
7915 rn[dim0] -= cg_cm[cg][i]*v_0[i][dim0];
7917 r2 = rn[dim0]*rn[dim0]*sf2_round[dim0];
7920 rb[dim0] = rn[dim0];
7923 /* Take care that the cell planes along dim0 might not
7924 * be orthogonal to those along dim1 and dim2.
7926 for (i = 1; i <= dim_ind; i++)
7929 if (normal[dim0][dimd] > 0)
7931 rn[dimd] -= rn[dim0]*normal[dim0][dimd];
7934 rb[dimd] -= rb[dim0]*normal[dim0][dimd];
7939 if (dim_ind == 2 && (zonei == 2 || zonei == 3))
7941 rn[dim1] += cg_cm[cg][dim1] - c->cr1[zone];
7943 for (i = dim1+1; i < DIM; i++)
7945 tric_sh -= cg_cm[cg][i]*v_1[i][dim1];
7947 rn[dim1] += tric_sh;
7950 r2 += rn[dim1]*rn[dim1]*sf2_round[dim1];
7951 /* Take care of coupling of the distances
7952 * to the planes along dim0 and dim1 through dim2.
7954 r2 -= rn[dim0]*rn[dim1]*skew_fac_01;
7955 /* Take care that the cell planes along dim1
7956 * might not be orthogonal to that along dim2.
7958 if (normal[dim1][dim2] > 0)
7960 rn[dim2] -= rn[dim1]*normal[dim1][dim2];
7966 cg_cm[cg][dim1] - c->bcr1 + tric_sh;
7969 rb2 += rb[dim1]*rb[dim1]*sf2_round[dim1];
7970 /* Take care of coupling of the distances
7971 * to the planes along dim0 and dim1 through dim2.
7973 rb2 -= rb[dim0]*rb[dim1]*skew_fac_01;
7974 /* Take care that the cell planes along dim1
7975 * might not be orthogonal to that along dim2.
7977 if (normal[dim1][dim2] > 0)
7979 rb[dim2] -= rb[dim1]*normal[dim1][dim2];
7984 /* The distance along the communication direction */
7985 rn[dim] += cg_cm[cg][dim] - c->c[dim_ind][zone];
7987 for (i = dim+1; i < DIM; i++)
7989 tric_sh -= cg_cm[cg][i]*v_d[i][dim];
7994 r2 += rn[dim]*rn[dim]*skew_fac2_d;
7995 /* Take care of coupling of the distances
7996 * to the planes along dim0 and dim1 through dim2.
7998 if (dim_ind == 1 && zonei == 1)
8000 r2 -= rn[dim0]*rn[dim]*skew_fac_01;
8006 rb[dim] += cg_cm[cg][dim] - c->bc[dim_ind] + tric_sh;
8009 rb2 += rb[dim]*rb[dim]*skew_fac2_d;
8010 /* Take care of coupling of the distances
8011 * to the planes along dim0 and dim1 through dim2.
8013 if (dim_ind == 1 && zonei == 1)
8015 rb2 -= rb[dim0]*rb[dim]*skew_fac_01;
8023 ((bDistMB && rb2 < r_bcomm2) ||
8024 (bDist2B && r2 < r_bcomm2)) &&
8026 (GET_CGINFO_BOND_INTER(cginfo[cg]) &&
8027 missing_link(comm->cglink, index_gl[cg],
8030 /* Make an index to the local charge groups */
8031 if (nsend+1 > ind->nalloc)
8033 ind->nalloc = over_alloc_large(nsend+1);
8034 srenew(ind->index, ind->nalloc);
8036 if (nsend+1 > *ibuf_nalloc)
8038 *ibuf_nalloc = over_alloc_large(nsend+1);
8039 srenew(*ibuf, *ibuf_nalloc);
8041 ind->index[nsend] = cg;
8042 (*ibuf)[nsend] = index_gl[cg];
8044 vec_rvec_check_alloc(vbuf, nsend+1);
8046 if (dd->ci[dim] == 0)
8048 /* Correct cg_cm for pbc */
8049 rvec_add(cg_cm[cg], box[dim], vbuf->v[nsend]);
8052 vbuf->v[nsend][YY] = box[YY][YY] - vbuf->v[nsend][YY];
8053 vbuf->v[nsend][ZZ] = box[ZZ][ZZ] - vbuf->v[nsend][ZZ];
8058 copy_rvec(cg_cm[cg], vbuf->v[nsend]);
8061 nat += cgindex[cg+1] - cgindex[cg];
8067 *nsend_z_ptr = nsend_z;
8070 static void setup_dd_communication(gmx_domdec_t *dd,
8071 matrix box, gmx_ddbox_t *ddbox,
8072 t_forcerec *fr, t_state *state, rvec **f)
8074 int dim_ind, dim, dim0, dim1, dim2, dimd, p, nat_tot;
8075 int nzone, nzone_send, zone, zonei, cg0, cg1;
8076 int c, i, j, cg, cg_gl, nrcg;
8077 int *zone_cg_range, pos_cg, *index_gl, *cgindex, *recv_i;
8078 gmx_domdec_comm_t *comm;
8079 gmx_domdec_zones_t *zones;
8080 gmx_domdec_comm_dim_t *cd;
8081 gmx_domdec_ind_t *ind;
8082 cginfo_mb_t *cginfo_mb;
8083 gmx_bool bBondComm, bDist2B, bDistMB, bDistBonded;
8084 real r_mb, r_comm2, r_scomm2, r_bcomm2, r_0, r_1, r2inc, inv_ncg;
8085 dd_corners_t corners;
8087 rvec *cg_cm, *normal, *v_d, *v_0 = NULL, *v_1 = NULL, *recv_vr;
8088 real skew_fac2_d, skew_fac_01;
8095 fprintf(debug, "Setting up DD communication\n");
8100 switch (fr->cutoff_scheme)
8109 gmx_incons("unimplemented");
8113 for (dim_ind = 0; dim_ind < dd->ndim; dim_ind++)
8115 dim = dd->dim[dim_ind];
8117 /* Check if we need to use triclinic distances */
8118 tric_dist[dim_ind] = 0;
8119 for (i = 0; i <= dim_ind; i++)
8121 if (ddbox->tric_dir[dd->dim[i]])
8123 tric_dist[dim_ind] = 1;
8128 bBondComm = comm->bBondComm;
8130 /* Do we need to determine extra distances for multi-body bondeds? */
8131 bDistMB = (comm->bInterCGMultiBody && dd->bGridJump && dd->ndim > 1);
8133 /* Do we need to determine extra distances for only two-body bondeds? */
8134 bDist2B = (bBondComm && !bDistMB);
8136 r_comm2 = sqr(comm->cutoff);
8137 r_bcomm2 = sqr(comm->cutoff_mbody);
8141 fprintf(debug, "bBondComm %d, r_bc %f\n", bBondComm, sqrt(r_bcomm2));
8144 zones = &comm->zones;
8147 dim1 = (dd->ndim >= 2 ? dd->dim[1] : -1);
8148 dim2 = (dd->ndim >= 3 ? dd->dim[2] : -1);
8150 set_dd_corners(dd, dim0, dim1, dim2, bDistMB, &corners);
8152 /* Triclinic stuff */
8153 normal = ddbox->normal;
8157 v_0 = ddbox->v[dim0];
8158 if (ddbox->tric_dir[dim0] && ddbox->tric_dir[dim1])
8160 /* Determine the coupling coefficient for the distances
8161 * to the cell planes along dim0 and dim1 through dim2.
8162 * This is required for correct rounding.
8165 ddbox->v[dim0][dim1+1][dim0]*ddbox->v[dim1][dim1+1][dim1];
8168 fprintf(debug, "\nskew_fac_01 %f\n", skew_fac_01);
8174 v_1 = ddbox->v[dim1];
8177 zone_cg_range = zones->cg_range;
8178 index_gl = dd->index_gl;
8179 cgindex = dd->cgindex;
8180 cginfo_mb = fr->cginfo_mb;
8182 zone_cg_range[0] = 0;
8183 zone_cg_range[1] = dd->ncg_home;
8184 comm->zone_ncg1[0] = dd->ncg_home;
8185 pos_cg = dd->ncg_home;
8187 nat_tot = dd->nat_home;
8189 for (dim_ind = 0; dim_ind < dd->ndim; dim_ind++)
8191 dim = dd->dim[dim_ind];
8192 cd = &comm->cd[dim_ind];
8194 if (dim >= ddbox->npbcdim && dd->ci[dim] == 0)
8196 /* No pbc in this dimension, the first node should not comm. */
8204 v_d = ddbox->v[dim];
8205 skew_fac2_d = sqr(ddbox->skew_fac[dim]);
8207 cd->bInPlace = TRUE;
8208 for (p = 0; p < cd->np; p++)
8210 /* Only atoms communicated in the first pulse are used
8211 * for multi-body bonded interactions or for bBondComm.
8213 bDistBonded = ((bDistMB || bDist2B) && p == 0);
8218 for (zone = 0; zone < nzone_send; zone++)
8220 if (tric_dist[dim_ind] && dim_ind > 0)
8222 /* Determine slightly more optimized skew_fac's
8224 * This reduces the number of communicated atoms
8225 * by about 10% for 3D DD of rhombic dodecahedra.
8227 for (dimd = 0; dimd < dim; dimd++)
8229 sf2_round[dimd] = 1;
8230 if (ddbox->tric_dir[dimd])
8232 for (i = dd->dim[dimd]+1; i < DIM; i++)
8234 /* If we are shifted in dimension i
8235 * and the cell plane is tilted forward
8236 * in dimension i, skip this coupling.
8238 if (!(zones->shift[nzone+zone][i] &&
8239 ddbox->v[dimd][i][dimd] >= 0))
8242 sqr(ddbox->v[dimd][i][dimd]);
8245 sf2_round[dimd] = 1/sf2_round[dimd];
8250 zonei = zone_perm[dim_ind][zone];
8253 /* Here we permutate the zones to obtain a convenient order
8254 * for neighbor searching
8256 cg0 = zone_cg_range[zonei];
8257 cg1 = zone_cg_range[zonei+1];
8261 /* Look only at the cg's received in the previous grid pulse
8263 cg1 = zone_cg_range[nzone+zone+1];
8264 cg0 = cg1 - cd->ind[p-1].nrecv[zone];
8267 #pragma omp parallel for num_threads(comm->nth) schedule(static)
8268 for (th = 0; th < comm->nth; th++)
8270 gmx_domdec_ind_t *ind_p;
8271 int **ibuf_p, *ibuf_nalloc_p;
8273 int *nsend_p, *nat_p;
8279 /* Thread 0 writes in the comm buffers */
8281 ibuf_p = &comm->buf_int;
8282 ibuf_nalloc_p = &comm->nalloc_int;
8283 vbuf_p = &comm->vbuf;
8286 nsend_zone_p = &ind->nsend[zone];
8290 /* Other threads write into temp buffers */
8291 ind_p = &comm->dth[th].ind;
8292 ibuf_p = &comm->dth[th].ibuf;
8293 ibuf_nalloc_p = &comm->dth[th].ibuf_nalloc;
8294 vbuf_p = &comm->dth[th].vbuf;
8295 nsend_p = &comm->dth[th].nsend;
8296 nat_p = &comm->dth[th].nat;
8297 nsend_zone_p = &comm->dth[th].nsend_zone;
8299 comm->dth[th].nsend = 0;
8300 comm->dth[th].nat = 0;
8301 comm->dth[th].nsend_zone = 0;
8311 cg0_th = cg0 + ((cg1 - cg0)* th )/comm->nth;
8312 cg1_th = cg0 + ((cg1 - cg0)*(th+1))/comm->nth;
8315 /* Get the cg's for this pulse in this zone */
8316 get_zone_pulse_cgs(dd, zonei, zone, cg0_th, cg1_th,
8318 dim, dim_ind, dim0, dim1, dim2,
8321 normal, skew_fac2_d, skew_fac_01,
8322 v_d, v_0, v_1, &corners, sf2_round,
8323 bDistBonded, bBondComm,
8327 ibuf_p, ibuf_nalloc_p,
8333 /* Append data of threads>=1 to the communication buffers */
8334 for (th = 1; th < comm->nth; th++)
8336 dd_comm_setup_work_t *dth;
8339 dth = &comm->dth[th];
8341 ns1 = nsend + dth->nsend_zone;
8342 if (ns1 > ind->nalloc)
8344 ind->nalloc = over_alloc_dd(ns1);
8345 srenew(ind->index, ind->nalloc);
8347 if (ns1 > comm->nalloc_int)
8349 comm->nalloc_int = over_alloc_dd(ns1);
8350 srenew(comm->buf_int, comm->nalloc_int);
8352 if (ns1 > comm->vbuf.nalloc)
8354 comm->vbuf.nalloc = over_alloc_dd(ns1);
8355 srenew(comm->vbuf.v, comm->vbuf.nalloc);
8358 for (i = 0; i < dth->nsend_zone; i++)
8360 ind->index[nsend] = dth->ind.index[i];
8361 comm->buf_int[nsend] = dth->ibuf[i];
8362 copy_rvec(dth->vbuf.v[i],
8363 comm->vbuf.v[nsend]);
8367 ind->nsend[zone] += dth->nsend_zone;
8370 /* Clear the counts in case we do not have pbc */
8371 for (zone = nzone_send; zone < nzone; zone++)
8373 ind->nsend[zone] = 0;
8375 ind->nsend[nzone] = nsend;
8376 ind->nsend[nzone+1] = nat;
8377 /* Communicate the number of cg's and atoms to receive */
8378 dd_sendrecv_int(dd, dim_ind, dddirBackward,
8379 ind->nsend, nzone+2,
8380 ind->nrecv, nzone+2);
8382 /* The rvec buffer is also required for atom buffers of size nsend
8383 * in dd_move_x and dd_move_f.
8385 vec_rvec_check_alloc(&comm->vbuf, ind->nsend[nzone+1]);
8389 /* We can receive in place if only the last zone is not empty */
8390 for (zone = 0; zone < nzone-1; zone++)
8392 if (ind->nrecv[zone] > 0)
8394 cd->bInPlace = FALSE;
8399 /* The int buffer is only required here for the cg indices */
8400 if (ind->nrecv[nzone] > comm->nalloc_int2)
8402 comm->nalloc_int2 = over_alloc_dd(ind->nrecv[nzone]);
8403 srenew(comm->buf_int2, comm->nalloc_int2);
8405 /* The rvec buffer is also required for atom buffers
8406 * of size nrecv in dd_move_x and dd_move_f.
8408 i = max(cd->ind[0].nrecv[nzone+1], ind->nrecv[nzone+1]);
8409 vec_rvec_check_alloc(&comm->vbuf2, i);
8413 /* Make space for the global cg indices */
8414 if (pos_cg + ind->nrecv[nzone] > dd->cg_nalloc
8415 || dd->cg_nalloc == 0)
8417 dd->cg_nalloc = over_alloc_dd(pos_cg + ind->nrecv[nzone]);
8418 srenew(index_gl, dd->cg_nalloc);
8419 srenew(cgindex, dd->cg_nalloc+1);
8421 /* Communicate the global cg indices */
8424 recv_i = index_gl + pos_cg;
8428 recv_i = comm->buf_int2;
8430 dd_sendrecv_int(dd, dim_ind, dddirBackward,
8431 comm->buf_int, nsend,
8432 recv_i, ind->nrecv[nzone]);
8434 /* Make space for cg_cm */
8435 dd_check_alloc_ncg(fr, state, f, pos_cg + ind->nrecv[nzone]);
8436 if (fr->cutoff_scheme == ecutsGROUP)
8444 /* Communicate cg_cm */
8447 recv_vr = cg_cm + pos_cg;
8451 recv_vr = comm->vbuf2.v;
8453 dd_sendrecv_rvec(dd, dim_ind, dddirBackward,
8454 comm->vbuf.v, nsend,
8455 recv_vr, ind->nrecv[nzone]);
8457 /* Make the charge group index */
8460 zone = (p == 0 ? 0 : nzone - 1);
8461 while (zone < nzone)
8463 for (cg = 0; cg < ind->nrecv[zone]; cg++)
8465 cg_gl = index_gl[pos_cg];
8466 fr->cginfo[pos_cg] = ddcginfo(cginfo_mb, cg_gl);
8467 nrcg = GET_CGINFO_NATOMS(fr->cginfo[pos_cg]);
8468 cgindex[pos_cg+1] = cgindex[pos_cg] + nrcg;
8471 /* Update the charge group presence,
8472 * so we can use it in the next pass of the loop.
8474 comm->bLocalCG[cg_gl] = TRUE;
8480 comm->zone_ncg1[nzone+zone] = ind->nrecv[zone];
8483 zone_cg_range[nzone+zone] = pos_cg;
8488 /* This part of the code is never executed with bBondComm. */
8489 merge_cg_buffers(nzone, cd, p, zone_cg_range,
8490 index_gl, recv_i, cg_cm, recv_vr,
8491 cgindex, fr->cginfo_mb, fr->cginfo);
8492 pos_cg += ind->nrecv[nzone];
8494 nat_tot += ind->nrecv[nzone+1];
8498 /* Store the atom block for easy copying of communication buffers */
8499 make_cell2at_index(cd, nzone, zone_cg_range[nzone], cgindex);
8503 dd->index_gl = index_gl;
8504 dd->cgindex = cgindex;
8506 dd->ncg_tot = zone_cg_range[zones->n];
8507 dd->nat_tot = nat_tot;
8508 comm->nat[ddnatHOME] = dd->nat_home;
8509 for (i = ddnatZONE; i < ddnatNR; i++)
8511 comm->nat[i] = dd->nat_tot;
8516 /* We don't need to update cginfo, since that was alrady done above.
8517 * So we pass NULL for the forcerec.
8519 dd_set_cginfo(dd->index_gl, dd->ncg_home, dd->ncg_tot,
8520 NULL, comm->bLocalCG);
8525 fprintf(debug, "Finished setting up DD communication, zones:");
8526 for (c = 0; c < zones->n; c++)
8528 fprintf(debug, " %d", zones->cg_range[c+1]-zones->cg_range[c]);
8530 fprintf(debug, "\n");
8534 static void set_cg_boundaries(gmx_domdec_zones_t *zones)
8538 for (c = 0; c < zones->nizone; c++)
8540 zones->izone[c].cg1 = zones->cg_range[c+1];
8541 zones->izone[c].jcg0 = zones->cg_range[zones->izone[c].j0];
8542 zones->izone[c].jcg1 = zones->cg_range[zones->izone[c].j1];
8546 static void set_zones_size(gmx_domdec_t *dd,
8547 matrix box, const gmx_ddbox_t *ddbox,
8548 int zone_start, int zone_end)
8550 gmx_domdec_comm_t *comm;
8551 gmx_domdec_zones_t *zones;
8553 int z, zi, zj0, zj1, d, dim;
8556 real size_j, add_tric;
8561 zones = &comm->zones;
8563 /* Do we need to determine extra distances for multi-body bondeds? */
8564 bDistMB = (comm->bInterCGMultiBody && dd->bGridJump && dd->ndim > 1);
8566 for (z = zone_start; z < zone_end; z++)
8568 /* Copy cell limits to zone limits.
8569 * Valid for non-DD dims and non-shifted dims.
8571 copy_rvec(comm->cell_x0, zones->size[z].x0);
8572 copy_rvec(comm->cell_x1, zones->size[z].x1);
8575 for (d = 0; d < dd->ndim; d++)
8579 for (z = 0; z < zones->n; z++)
8581 /* With a staggered grid we have different sizes
8582 * for non-shifted dimensions.
8584 if (dd->bGridJump && zones->shift[z][dim] == 0)
8588 zones->size[z].x0[dim] = comm->zone_d1[zones->shift[z][dd->dim[d-1]]].min0;
8589 zones->size[z].x1[dim] = comm->zone_d1[zones->shift[z][dd->dim[d-1]]].max1;
8593 zones->size[z].x0[dim] = comm->zone_d2[zones->shift[z][dd->dim[d-2]]][zones->shift[z][dd->dim[d-1]]].min0;
8594 zones->size[z].x1[dim] = comm->zone_d2[zones->shift[z][dd->dim[d-2]]][zones->shift[z][dd->dim[d-1]]].max1;
8600 rcmbs = comm->cutoff_mbody;
8601 if (ddbox->tric_dir[dim])
8603 rcs /= ddbox->skew_fac[dim];
8604 rcmbs /= ddbox->skew_fac[dim];
8607 /* Set the lower limit for the shifted zone dimensions */
8608 for (z = zone_start; z < zone_end; z++)
8610 if (zones->shift[z][dim] > 0)
8613 if (!dd->bGridJump || d == 0)
8615 zones->size[z].x0[dim] = comm->cell_x1[dim];
8616 zones->size[z].x1[dim] = comm->cell_x1[dim] + rcs;
8620 /* Here we take the lower limit of the zone from
8621 * the lowest domain of the zone below.
8625 zones->size[z].x0[dim] =
8626 comm->zone_d1[zones->shift[z][dd->dim[d-1]]].min1;
8632 zones->size[z].x0[dim] =
8633 zones->size[zone_perm[2][z-4]].x0[dim];
8637 zones->size[z].x0[dim] =
8638 comm->zone_d2[zones->shift[z][dd->dim[d-2]]][zones->shift[z][dd->dim[d-1]]].min1;
8641 /* A temporary limit, is updated below */
8642 zones->size[z].x1[dim] = zones->size[z].x0[dim];
8646 for (zi = 0; zi < zones->nizone; zi++)
8648 if (zones->shift[zi][dim] == 0)
8650 /* This takes the whole zone into account.
8651 * With multiple pulses this will lead
8652 * to a larger zone then strictly necessary.
8654 zones->size[z].x1[dim] = max(zones->size[z].x1[dim],
8655 zones->size[zi].x1[dim]+rcmbs);
8663 /* Loop over the i-zones to set the upper limit of each
8666 for (zi = 0; zi < zones->nizone; zi++)
8668 if (zones->shift[zi][dim] == 0)
8670 for (z = zones->izone[zi].j0; z < zones->izone[zi].j1; z++)
8672 if (zones->shift[z][dim] > 0)
8674 zones->size[z].x1[dim] = max(zones->size[z].x1[dim],
8675 zones->size[zi].x1[dim]+rcs);
8682 for (z = zone_start; z < zone_end; z++)
8684 /* Initialization only required to keep the compiler happy */
8685 rvec corner_min = {0, 0, 0}, corner_max = {0, 0, 0}, corner;
8688 /* To determine the bounding box for a zone we need to find
8689 * the extreme corners of 4, 2 or 1 corners.
8691 nc = 1 << (ddbox->npbcdim - 1);
8693 for (c = 0; c < nc; c++)
8695 /* Set up a zone corner at x=0, ignoring trilinic couplings */
8699 corner[YY] = zones->size[z].x0[YY];
8703 corner[YY] = zones->size[z].x1[YY];
8707 corner[ZZ] = zones->size[z].x0[ZZ];
8711 corner[ZZ] = zones->size[z].x1[ZZ];
8713 if (dd->ndim == 1 && box[ZZ][YY] != 0)
8715 /* With 1D domain decomposition the cg's are not in
8716 * the triclinic box, but triclinic x-y and rectangular y-z.
8717 * Shift y back, so it will later end up at 0.
8719 corner[YY] -= corner[ZZ]*box[ZZ][YY]/box[ZZ][ZZ];
8721 /* Apply the triclinic couplings */
8722 assert(ddbox->npbcdim <= DIM);
8723 for (i = YY; i < ddbox->npbcdim; i++)
8725 for (j = XX; j < i; j++)
8727 corner[j] += corner[i]*box[i][j]/box[i][i];
8732 copy_rvec(corner, corner_min);
8733 copy_rvec(corner, corner_max);
8737 for (i = 0; i < DIM; i++)
8739 corner_min[i] = min(corner_min[i], corner[i]);
8740 corner_max[i] = max(corner_max[i], corner[i]);
8744 /* Copy the extreme cornes without offset along x */
8745 for (i = 0; i < DIM; i++)
8747 zones->size[z].bb_x0[i] = corner_min[i];
8748 zones->size[z].bb_x1[i] = corner_max[i];
8750 /* Add the offset along x */
8751 zones->size[z].bb_x0[XX] += zones->size[z].x0[XX];
8752 zones->size[z].bb_x1[XX] += zones->size[z].x1[XX];
8755 if (zone_start == 0)
8758 for (dim = 0; dim < DIM; dim++)
8760 vol *= zones->size[0].x1[dim] - zones->size[0].x0[dim];
8762 zones->dens_zone0 = (zones->cg_range[1] - zones->cg_range[0])/vol;
8767 for (z = zone_start; z < zone_end; z++)
8769 fprintf(debug, "zone %d %6.3f - %6.3f %6.3f - %6.3f %6.3f - %6.3f\n",
8771 zones->size[z].x0[XX], zones->size[z].x1[XX],
8772 zones->size[z].x0[YY], zones->size[z].x1[YY],
8773 zones->size[z].x0[ZZ], zones->size[z].x1[ZZ]);
8774 fprintf(debug, "zone %d bb %6.3f - %6.3f %6.3f - %6.3f %6.3f - %6.3f\n",
8776 zones->size[z].bb_x0[XX], zones->size[z].bb_x1[XX],
8777 zones->size[z].bb_x0[YY], zones->size[z].bb_x1[YY],
8778 zones->size[z].bb_x0[ZZ], zones->size[z].bb_x1[ZZ]);
8783 static int comp_cgsort(const void *a, const void *b)
8787 gmx_cgsort_t *cga, *cgb;
8788 cga = (gmx_cgsort_t *)a;
8789 cgb = (gmx_cgsort_t *)b;
8791 comp = cga->nsc - cgb->nsc;
8794 comp = cga->ind_gl - cgb->ind_gl;
8800 static void order_int_cg(int n, const gmx_cgsort_t *sort,
8805 /* Order the data */
8806 for (i = 0; i < n; i++)
8808 buf[i] = a[sort[i].ind];
8811 /* Copy back to the original array */
8812 for (i = 0; i < n; i++)
8818 static void order_vec_cg(int n, const gmx_cgsort_t *sort,
8823 /* Order the data */
8824 for (i = 0; i < n; i++)
8826 copy_rvec(v[sort[i].ind], buf[i]);
8829 /* Copy back to the original array */
8830 for (i = 0; i < n; i++)
8832 copy_rvec(buf[i], v[i]);
8836 static void order_vec_atom(int ncg, const int *cgindex, const gmx_cgsort_t *sort,
8839 int a, atot, cg, cg0, cg1, i;
8841 if (cgindex == NULL)
8843 /* Avoid the useless loop of the atoms within a cg */
8844 order_vec_cg(ncg, sort, v, buf);
8849 /* Order the data */
8851 for (cg = 0; cg < ncg; cg++)
8853 cg0 = cgindex[sort[cg].ind];
8854 cg1 = cgindex[sort[cg].ind+1];
8855 for (i = cg0; i < cg1; i++)
8857 copy_rvec(v[i], buf[a]);
8863 /* Copy back to the original array */
8864 for (a = 0; a < atot; a++)
8866 copy_rvec(buf[a], v[a]);
8870 static void ordered_sort(int nsort2, gmx_cgsort_t *sort2,
8871 int nsort_new, gmx_cgsort_t *sort_new,
8872 gmx_cgsort_t *sort1)
8876 /* The new indices are not very ordered, so we qsort them */
8877 gmx_qsort_threadsafe(sort_new, nsort_new, sizeof(sort_new[0]), comp_cgsort);
8879 /* sort2 is already ordered, so now we can merge the two arrays */
8883 while (i2 < nsort2 || i_new < nsort_new)
8887 sort1[i1++] = sort_new[i_new++];
8889 else if (i_new == nsort_new)
8891 sort1[i1++] = sort2[i2++];
8893 else if (sort2[i2].nsc < sort_new[i_new].nsc ||
8894 (sort2[i2].nsc == sort_new[i_new].nsc &&
8895 sort2[i2].ind_gl < sort_new[i_new].ind_gl))
8897 sort1[i1++] = sort2[i2++];
8901 sort1[i1++] = sort_new[i_new++];
8906 static int dd_sort_order(gmx_domdec_t *dd, t_forcerec *fr, int ncg_home_old)
8908 gmx_domdec_sort_t *sort;
8909 gmx_cgsort_t *cgsort, *sort_i;
8910 int ncg_new, nsort2, nsort_new, i, *a, moved, *ibuf;
8911 int sort_last, sort_skip;
8913 sort = dd->comm->sort;
8915 a = fr->ns.grid->cell_index;
8917 moved = NSGRID_SIGNAL_MOVED_FAC*fr->ns.grid->ncells;
8919 if (ncg_home_old >= 0)
8921 /* The charge groups that remained in the same ns grid cell
8922 * are completely ordered. So we can sort efficiently by sorting
8923 * the charge groups that did move into the stationary list.
8928 for (i = 0; i < dd->ncg_home; i++)
8930 /* Check if this cg did not move to another node */
8933 if (i >= ncg_home_old || a[i] != sort->sort[i].nsc)
8935 /* This cg is new on this node or moved ns grid cell */
8936 if (nsort_new >= sort->sort_new_nalloc)
8938 sort->sort_new_nalloc = over_alloc_dd(nsort_new+1);
8939 srenew(sort->sort_new, sort->sort_new_nalloc);
8941 sort_i = &(sort->sort_new[nsort_new++]);
8945 /* This cg did not move */
8946 sort_i = &(sort->sort2[nsort2++]);
8948 /* Sort on the ns grid cell indices
8949 * and the global topology index.
8950 * index_gl is irrelevant with cell ns,
8951 * but we set it here anyhow to avoid a conditional.
8954 sort_i->ind_gl = dd->index_gl[i];
8961 fprintf(debug, "ordered sort cgs: stationary %d moved %d\n",
8964 /* Sort efficiently */
8965 ordered_sort(nsort2, sort->sort2, nsort_new, sort->sort_new,
8970 cgsort = sort->sort;
8972 for (i = 0; i < dd->ncg_home; i++)
8974 /* Sort on the ns grid cell indices
8975 * and the global topology index
8977 cgsort[i].nsc = a[i];
8978 cgsort[i].ind_gl = dd->index_gl[i];
8980 if (cgsort[i].nsc < moved)
8987 fprintf(debug, "qsort cgs: %d new home %d\n", dd->ncg_home, ncg_new);
8989 /* Determine the order of the charge groups using qsort */
8990 gmx_qsort_threadsafe(cgsort, dd->ncg_home, sizeof(cgsort[0]), comp_cgsort);
8996 static int dd_sort_order_nbnxn(gmx_domdec_t *dd, t_forcerec *fr)
8999 int ncg_new, i, *a, na;
9001 sort = dd->comm->sort->sort;
9003 nbnxn_get_atomorder(fr->nbv->nbs, &a, &na);
9006 for (i = 0; i < na; i++)
9010 sort[ncg_new].ind = a[i];
9018 static void dd_sort_state(gmx_domdec_t *dd, rvec *cgcm, t_forcerec *fr, t_state *state,
9021 gmx_domdec_sort_t *sort;
9022 gmx_cgsort_t *cgsort, *sort_i;
9024 int ncg_new, i, *ibuf, cgsize;
9027 sort = dd->comm->sort;
9029 if (dd->ncg_home > sort->sort_nalloc)
9031 sort->sort_nalloc = over_alloc_dd(dd->ncg_home);
9032 srenew(sort->sort, sort->sort_nalloc);
9033 srenew(sort->sort2, sort->sort_nalloc);
9035 cgsort = sort->sort;
9037 switch (fr->cutoff_scheme)
9040 ncg_new = dd_sort_order(dd, fr, ncg_home_old);
9043 ncg_new = dd_sort_order_nbnxn(dd, fr);
9046 gmx_incons("unimplemented");
9050 /* We alloc with the old size, since cgindex is still old */
9051 vec_rvec_check_alloc(&dd->comm->vbuf, dd->cgindex[dd->ncg_home]);
9052 vbuf = dd->comm->vbuf.v;
9056 cgindex = dd->cgindex;
9063 /* Remove the charge groups which are no longer at home here */
9064 dd->ncg_home = ncg_new;
9067 fprintf(debug, "Set the new home charge group count to %d\n",
9071 /* Reorder the state */
9072 for (i = 0; i < estNR; i++)
9074 if (EST_DISTR(i) && (state->flags & (1<<i)))
9079 order_vec_atom(dd->ncg_home, cgindex, cgsort, state->x, vbuf);
9082 order_vec_atom(dd->ncg_home, cgindex, cgsort, state->v, vbuf);
9085 order_vec_atom(dd->ncg_home, cgindex, cgsort, state->sd_X, vbuf);
9088 order_vec_atom(dd->ncg_home, cgindex, cgsort, state->cg_p, vbuf);
9092 case estDISRE_INITF:
9093 case estDISRE_RM3TAV:
9094 case estORIRE_INITF:
9096 /* No ordering required */
9099 gmx_incons("Unknown state entry encountered in dd_sort_state");
9104 if (fr->cutoff_scheme == ecutsGROUP)
9107 order_vec_cg(dd->ncg_home, cgsort, cgcm, vbuf);
9110 if (dd->ncg_home+1 > sort->ibuf_nalloc)
9112 sort->ibuf_nalloc = over_alloc_dd(dd->ncg_home+1);
9113 srenew(sort->ibuf, sort->ibuf_nalloc);
9116 /* Reorder the global cg index */
9117 order_int_cg(dd->ncg_home, cgsort, dd->index_gl, ibuf);
9118 /* Reorder the cginfo */
9119 order_int_cg(dd->ncg_home, cgsort, fr->cginfo, ibuf);
9120 /* Rebuild the local cg index */
9124 for (i = 0; i < dd->ncg_home; i++)
9126 cgsize = dd->cgindex[cgsort[i].ind+1] - dd->cgindex[cgsort[i].ind];
9127 ibuf[i+1] = ibuf[i] + cgsize;
9129 for (i = 0; i < dd->ncg_home+1; i++)
9131 dd->cgindex[i] = ibuf[i];
9136 for (i = 0; i < dd->ncg_home+1; i++)
9141 /* Set the home atom number */
9142 dd->nat_home = dd->cgindex[dd->ncg_home];
9144 if (fr->cutoff_scheme == ecutsVERLET)
9146 /* The atoms are now exactly in grid order, update the grid order */
9147 nbnxn_set_atomorder(fr->nbv->nbs);
9151 /* Copy the sorted ns cell indices back to the ns grid struct */
9152 for (i = 0; i < dd->ncg_home; i++)
9154 fr->ns.grid->cell_index[i] = cgsort[i].nsc;
9156 fr->ns.grid->nr = dd->ncg_home;
9160 static void add_dd_statistics(gmx_domdec_t *dd)
9162 gmx_domdec_comm_t *comm;
9167 for (ddnat = ddnatZONE; ddnat < ddnatNR; ddnat++)
9169 comm->sum_nat[ddnat-ddnatZONE] +=
9170 comm->nat[ddnat] - comm->nat[ddnat-1];
9175 void reset_dd_statistics_counters(gmx_domdec_t *dd)
9177 gmx_domdec_comm_t *comm;
9182 /* Reset all the statistics and counters for total run counting */
9183 for (ddnat = ddnatZONE; ddnat < ddnatNR; ddnat++)
9185 comm->sum_nat[ddnat-ddnatZONE] = 0;
9189 comm->load_step = 0;
9192 clear_ivec(comm->load_lim);
9197 void print_dd_statistics(t_commrec *cr, t_inputrec *ir, FILE *fplog)
9199 gmx_domdec_comm_t *comm;
9203 comm = cr->dd->comm;
9205 gmx_sumd(ddnatNR-ddnatZONE, comm->sum_nat, cr);
9212 fprintf(fplog, "\n D O M A I N D E C O M P O S I T I O N S T A T I S T I C S\n\n");
9214 for (ddnat = ddnatZONE; ddnat < ddnatNR; ddnat++)
9216 av = comm->sum_nat[ddnat-ddnatZONE]/comm->ndecomp;
9221 " av. #atoms communicated per step for force: %d x %.1f\n",
9225 if (cr->dd->vsite_comm)
9228 " av. #atoms communicated per step for vsites: %d x %.1f\n",
9229 (EEL_PME(ir->coulombtype) || ir->coulombtype == eelEWALD) ? 3 : 2,
9234 if (cr->dd->constraint_comm)
9237 " av. #atoms communicated per step for LINCS: %d x %.1f\n",
9238 1 + ir->nLincsIter, av);
9242 gmx_incons(" Unknown type for DD statistics");
9245 fprintf(fplog, "\n");
9247 if (comm->bRecordLoad && EI_DYNAMICS(ir->eI))
9249 print_dd_load_av(fplog, cr->dd);
9253 void dd_partition_system(FILE *fplog,
9256 gmx_bool bMasterState,
9258 t_state *state_global,
9259 gmx_mtop_t *top_global,
9261 t_state *state_local,
9264 gmx_localtop_t *top_local,
9267 gmx_shellfc_t shellfc,
9268 gmx_constr_t constr,
9270 gmx_wallcycle_t wcycle,
9274 gmx_domdec_comm_t *comm;
9275 gmx_ddbox_t ddbox = {0};
9277 gmx_int64_t step_pcoupl;
9278 rvec cell_ns_x0, cell_ns_x1;
9279 int i, j, n, ncgindex_set, ncg_home_old = -1, ncg_moved, nat_f_novirsum;
9280 gmx_bool bBoxChanged, bNStGlobalComm, bDoDLB, bCheckDLB, bTurnOnDLB, bLogLoad;
9281 gmx_bool bRedist, bSortCG, bResortAll;
9282 ivec ncells_old = {0, 0, 0}, ncells_new = {0, 0, 0}, np;
9289 bBoxChanged = (bMasterState || DEFORM(*ir));
9290 if (ir->epc != epcNO)
9292 /* With nstpcouple > 1 pressure coupling happens.
9293 * one step after calculating the pressure.
9294 * Box scaling happens at the end of the MD step,
9295 * after the DD partitioning.
9296 * We therefore have to do DLB in the first partitioning
9297 * after an MD step where P-coupling occured.
9298 * We need to determine the last step in which p-coupling occurred.
9299 * MRS -- need to validate this for vv?
9304 step_pcoupl = step - 1;
9308 step_pcoupl = ((step - 1)/n)*n + 1;
9310 if (step_pcoupl >= comm->partition_step)
9316 bNStGlobalComm = (step % nstglobalcomm == 0);
9318 if (!comm->bDynLoadBal)
9324 /* Should we do dynamic load balacing this step?
9325 * Since it requires (possibly expensive) global communication,
9326 * we might want to do DLB less frequently.
9328 if (bBoxChanged || ir->epc != epcNO)
9330 bDoDLB = bBoxChanged;
9334 bDoDLB = bNStGlobalComm;
9338 /* Check if we have recorded loads on the nodes */
9339 if (comm->bRecordLoad && dd_load_count(comm))
9341 if (comm->eDLB == edlbAUTO && !comm->bDynLoadBal)
9343 /* Check if we should use DLB at the second partitioning
9344 * and every 100 partitionings,
9345 * so the extra communication cost is negligible.
9347 n = max(100, nstglobalcomm);
9348 bCheckDLB = (comm->n_load_collect == 0 ||
9349 comm->n_load_have % n == n-1);
9356 /* Print load every nstlog, first and last step to the log file */
9357 bLogLoad = ((ir->nstlog > 0 && step % ir->nstlog == 0) ||
9358 comm->n_load_collect == 0 ||
9360 (step + ir->nstlist > ir->init_step + ir->nsteps)));
9362 /* Avoid extra communication due to verbose screen output
9363 * when nstglobalcomm is set.
9365 if (bDoDLB || bLogLoad || bCheckDLB ||
9366 (bVerbose && (ir->nstlist == 0 || nstglobalcomm <= ir->nstlist)))
9368 get_load_distribution(dd, wcycle);
9373 dd_print_load(fplog, dd, step-1);
9377 dd_print_load_verbose(dd);
9380 comm->n_load_collect++;
9384 /* Since the timings are node dependent, the master decides */
9388 (dd_force_imb_perf_loss(dd) >= DD_PERF_LOSS_DLB_ON);
9391 fprintf(debug, "step %s, imb loss %f\n",
9392 gmx_step_str(step, sbuf),
9393 dd_force_imb_perf_loss(dd));
9396 dd_bcast(dd, sizeof(bTurnOnDLB), &bTurnOnDLB);
9399 turn_on_dlb(fplog, cr, step);
9404 comm->n_load_have++;
9407 cgs_gl = &comm->cgs_gl;
9412 /* Clear the old state */
9413 clear_dd_indices(dd, 0, 0);
9416 set_ddbox(dd, bMasterState, cr, ir, state_global->box,
9417 TRUE, cgs_gl, state_global->x, &ddbox);
9419 get_cg_distribution(fplog, step, dd, cgs_gl,
9420 state_global->box, &ddbox, state_global->x);
9422 dd_distribute_state(dd, cgs_gl,
9423 state_global, state_local, f);
9425 dd_make_local_cgs(dd, &top_local->cgs);
9427 /* Ensure that we have space for the new distribution */
9428 dd_check_alloc_ncg(fr, state_local, f, dd->ncg_home);
9430 if (fr->cutoff_scheme == ecutsGROUP)
9432 calc_cgcm(fplog, 0, dd->ncg_home,
9433 &top_local->cgs, state_local->x, fr->cg_cm);
9436 inc_nrnb(nrnb, eNR_CGCM, dd->nat_home);
9438 dd_set_cginfo(dd->index_gl, 0, dd->ncg_home, fr, comm->bLocalCG);
9440 else if (state_local->ddp_count != dd->ddp_count)
9442 if (state_local->ddp_count > dd->ddp_count)
9444 gmx_fatal(FARGS, "Internal inconsistency state_local->ddp_count (%d) > dd->ddp_count (%d)", state_local->ddp_count, dd->ddp_count);
9447 if (state_local->ddp_count_cg_gl != state_local->ddp_count)
9449 gmx_fatal(FARGS, "Internal inconsistency state_local->ddp_count_cg_gl (%d) != state_local->ddp_count (%d)", state_local->ddp_count_cg_gl, state_local->ddp_count);
9452 /* Clear the old state */
9453 clear_dd_indices(dd, 0, 0);
9455 /* Build the new indices */
9456 rebuild_cgindex(dd, cgs_gl->index, state_local);
9457 make_dd_indices(dd, cgs_gl->index, 0);
9458 ncgindex_set = dd->ncg_home;
9460 if (fr->cutoff_scheme == ecutsGROUP)
9462 /* Redetermine the cg COMs */
9463 calc_cgcm(fplog, 0, dd->ncg_home,
9464 &top_local->cgs, state_local->x, fr->cg_cm);
9467 inc_nrnb(nrnb, eNR_CGCM, dd->nat_home);
9469 dd_set_cginfo(dd->index_gl, 0, dd->ncg_home, fr, comm->bLocalCG);
9471 set_ddbox(dd, bMasterState, cr, ir, state_local->box,
9472 TRUE, &top_local->cgs, state_local->x, &ddbox);
9474 bRedist = comm->bDynLoadBal;
9478 /* We have the full state, only redistribute the cgs */
9480 /* Clear the non-home indices */
9481 clear_dd_indices(dd, dd->ncg_home, dd->nat_home);
9484 /* Avoid global communication for dim's without pbc and -gcom */
9485 if (!bNStGlobalComm)
9487 copy_rvec(comm->box0, ddbox.box0 );
9488 copy_rvec(comm->box_size, ddbox.box_size);
9490 set_ddbox(dd, bMasterState, cr, ir, state_local->box,
9491 bNStGlobalComm, &top_local->cgs, state_local->x, &ddbox);
9496 /* For dim's without pbc and -gcom */
9497 copy_rvec(ddbox.box0, comm->box0 );
9498 copy_rvec(ddbox.box_size, comm->box_size);
9500 set_dd_cell_sizes(dd, &ddbox, dynamic_dd_box(&ddbox, ir), bMasterState, bDoDLB,
9503 if (comm->nstDDDumpGrid > 0 && step % comm->nstDDDumpGrid == 0)
9505 write_dd_grid_pdb("dd_grid", step, dd, state_local->box, &ddbox);
9508 /* Check if we should sort the charge groups */
9509 if (comm->nstSortCG > 0)
9511 bSortCG = (bMasterState ||
9512 (bRedist && (step % comm->nstSortCG == 0)));
9519 ncg_home_old = dd->ncg_home;
9524 wallcycle_sub_start(wcycle, ewcsDD_REDIST);
9526 dd_redistribute_cg(fplog, step, dd, ddbox.tric_dir,
9528 !bSortCG, nrnb, &ncgindex_set, &ncg_moved);
9530 wallcycle_sub_stop(wcycle, ewcsDD_REDIST);
9533 get_nsgrid_boundaries(ddbox.nboundeddim, state_local->box,
9535 &comm->cell_x0, &comm->cell_x1,
9536 dd->ncg_home, fr->cg_cm,
9537 cell_ns_x0, cell_ns_x1, &grid_density);
9541 comm_dd_ns_cell_sizes(dd, &ddbox, cell_ns_x0, cell_ns_x1, step);
9544 switch (fr->cutoff_scheme)
9547 copy_ivec(fr->ns.grid->n, ncells_old);
9548 grid_first(fplog, fr->ns.grid, dd, &ddbox,
9549 state_local->box, cell_ns_x0, cell_ns_x1,
9550 fr->rlistlong, grid_density);
9553 nbnxn_get_ncells(fr->nbv->nbs, &ncells_old[XX], &ncells_old[YY]);
9556 gmx_incons("unimplemented");
9558 /* We need to store tric_dir for dd_get_ns_ranges called from ns.c */
9559 copy_ivec(ddbox.tric_dir, comm->tric_dir);
9563 wallcycle_sub_start(wcycle, ewcsDD_GRID);
9565 /* Sort the state on charge group position.
9566 * This enables exact restarts from this step.
9567 * It also improves performance by about 15% with larger numbers
9568 * of atoms per node.
9571 /* Fill the ns grid with the home cell,
9572 * so we can sort with the indices.
9574 set_zones_ncg_home(dd);
9576 switch (fr->cutoff_scheme)
9579 set_zones_size(dd, state_local->box, &ddbox, 0, 1);
9581 nbnxn_put_on_grid(fr->nbv->nbs, fr->ePBC, state_local->box,
9583 comm->zones.size[0].bb_x0,
9584 comm->zones.size[0].bb_x1,
9586 comm->zones.dens_zone0,
9589 ncg_moved, bRedist ? comm->moved : NULL,
9590 fr->nbv->grp[eintLocal].kernel_type,
9591 fr->nbv->grp[eintLocal].nbat);
9593 nbnxn_get_ncells(fr->nbv->nbs, &ncells_new[XX], &ncells_new[YY]);
9596 fill_grid(&comm->zones, fr->ns.grid, dd->ncg_home,
9597 0, dd->ncg_home, fr->cg_cm);
9599 copy_ivec(fr->ns.grid->n, ncells_new);
9602 gmx_incons("unimplemented");
9605 bResortAll = bMasterState;
9607 /* Check if we can user the old order and ns grid cell indices
9608 * of the charge groups to sort the charge groups efficiently.
9610 if (ncells_new[XX] != ncells_old[XX] ||
9611 ncells_new[YY] != ncells_old[YY] ||
9612 ncells_new[ZZ] != ncells_old[ZZ])
9619 fprintf(debug, "Step %s, sorting the %d home charge groups\n",
9620 gmx_step_str(step, sbuf), dd->ncg_home);
9622 dd_sort_state(dd, fr->cg_cm, fr, state_local,
9623 bResortAll ? -1 : ncg_home_old);
9624 /* Rebuild all the indices */
9625 ga2la_clear(dd->ga2la);
9628 wallcycle_sub_stop(wcycle, ewcsDD_GRID);
9631 wallcycle_sub_start(wcycle, ewcsDD_SETUPCOMM);
9633 /* Setup up the communication and communicate the coordinates */
9634 setup_dd_communication(dd, state_local->box, &ddbox, fr, state_local, f);
9636 /* Set the indices */
9637 make_dd_indices(dd, cgs_gl->index, ncgindex_set);
9639 /* Set the charge group boundaries for neighbor searching */
9640 set_cg_boundaries(&comm->zones);
9642 if (fr->cutoff_scheme == ecutsVERLET)
9644 set_zones_size(dd, state_local->box, &ddbox,
9645 bSortCG ? 1 : 0, comm->zones.n);
9648 wallcycle_sub_stop(wcycle, ewcsDD_SETUPCOMM);
9651 write_dd_pdb("dd_home",step,"dump",top_global,cr,
9652 -1,state_local->x,state_local->box);
9655 wallcycle_sub_start(wcycle, ewcsDD_MAKETOP);
9657 /* Extract a local topology from the global topology */
9658 for (i = 0; i < dd->ndim; i++)
9660 np[dd->dim[i]] = comm->cd[i].np;
9662 dd_make_local_top(dd, &comm->zones, dd->npbcdim, state_local->box,
9663 comm->cellsize_min, np,
9665 fr->cutoff_scheme == ecutsGROUP ? fr->cg_cm : state_local->x,
9666 vsite, top_global, top_local);
9668 wallcycle_sub_stop(wcycle, ewcsDD_MAKETOP);
9670 wallcycle_sub_start(wcycle, ewcsDD_MAKECONSTR);
9672 /* Set up the special atom communication */
9673 n = comm->nat[ddnatZONE];
9674 for (i = ddnatZONE+1; i < ddnatNR; i++)
9679 if (vsite && vsite->n_intercg_vsite)
9681 n = dd_make_local_vsites(dd, n, top_local->idef.il);
9685 if (dd->bInterCGcons || dd->bInterCGsettles)
9687 /* Only for inter-cg constraints we need special code */
9688 n = dd_make_local_constraints(dd, n, top_global, fr->cginfo,
9689 constr, ir->nProjOrder,
9690 top_local->idef.il);
9694 gmx_incons("Unknown special atom type setup");
9699 wallcycle_sub_stop(wcycle, ewcsDD_MAKECONSTR);
9701 wallcycle_sub_start(wcycle, ewcsDD_TOPOTHER);
9703 /* Make space for the extra coordinates for virtual site
9704 * or constraint communication.
9706 state_local->natoms = comm->nat[ddnatNR-1];
9707 if (state_local->natoms > state_local->nalloc)
9709 dd_realloc_state(state_local, f, state_local->natoms);
9712 if (fr->bF_NoVirSum)
9714 if (vsite && vsite->n_intercg_vsite)
9716 nat_f_novirsum = comm->nat[ddnatVSITE];
9720 if (EEL_FULL(ir->coulombtype) && dd->n_intercg_excl > 0)
9722 nat_f_novirsum = dd->nat_tot;
9726 nat_f_novirsum = dd->nat_home;
9735 /* Set the number of atoms required for the force calculation.
9736 * Forces need to be constrained when using a twin-range setup
9737 * or with energy minimization. For simple simulations we could
9738 * avoid some allocation, zeroing and copying, but this is
9739 * probably not worth the complications ande checking.
9741 forcerec_set_ranges(fr, dd->ncg_home, dd->ncg_tot,
9742 dd->nat_tot, comm->nat[ddnatCON], nat_f_novirsum);
9744 /* We make the all mdatoms up to nat_tot_con.
9745 * We could save some work by only setting invmass
9746 * between nat_tot and nat_tot_con.
9748 /* This call also sets the new number of home particles to dd->nat_home */
9749 atoms2md(top_global, ir,
9750 comm->nat[ddnatCON], dd->gatindex, dd->nat_home, mdatoms);
9752 /* Now we have the charges we can sort the FE interactions */
9753 dd_sort_local_top(dd, mdatoms, top_local);
9757 /* Now we have updated mdatoms, we can do the last vsite bookkeeping */
9758 split_vsites_over_threads(top_local->idef.il, top_local->idef.iparams,
9759 mdatoms, FALSE, vsite);
9764 /* Make the local shell stuff, currently no communication is done */
9765 make_local_shells(cr, mdatoms, shellfc);
9768 if (ir->implicit_solvent)
9770 make_local_gb(cr, fr->born, ir->gb_algorithm);
9773 setup_bonded_threading(fr, &top_local->idef);
9775 if (!(cr->duty & DUTY_PME))
9777 /* Send the charges and/or c6/sigmas to our PME only node */
9778 gmx_pme_send_parameters(cr,
9780 mdatoms->nChargePerturbed, mdatoms->nTypePerturbed,
9781 mdatoms->chargeA, mdatoms->chargeB,
9782 mdatoms->sqrt_c6A, mdatoms->sqrt_c6B,
9783 mdatoms->sigmaA, mdatoms->sigmaB,
9784 dd_pme_maxshift_x(dd), dd_pme_maxshift_y(dd));
9789 set_constraints(constr, top_local, ir, mdatoms, cr);
9792 if (ir->ePull != epullNO)
9794 /* Update the local pull groups */
9795 dd_make_local_pull_groups(dd, ir->pull, mdatoms);
9800 /* Update the local rotation groups */
9801 dd_make_local_rotation_groups(dd, ir->rot);
9804 if (ir->eSwapCoords != eswapNO)
9806 /* Update the local groups needed for ion swapping */
9807 dd_make_local_swap_groups(dd, ir->swap);
9810 /* Update the local atoms to be communicated via the IMD protocol if bIMD is TRUE. */
9811 dd_make_local_IMD_atoms(ir->bIMD, dd, ir->imd);
9813 add_dd_statistics(dd);
9815 /* Make sure we only count the cycles for this DD partitioning */
9816 clear_dd_cycle_counts(dd);
9818 /* Because the order of the atoms might have changed since
9819 * the last vsite construction, we need to communicate the constructing
9820 * atom coordinates again (for spreading the forces this MD step).
9822 dd_move_x_vsites(dd, state_local->box, state_local->x);
9824 wallcycle_sub_stop(wcycle, ewcsDD_TOPOTHER);
9826 if (comm->nstDDDump > 0 && step % comm->nstDDDump == 0)
9828 dd_move_x(dd, state_local->box, state_local->x);
9829 write_dd_pdb("dd_dump", step, "dump", top_global, cr,
9830 -1, state_local->x, state_local->box);
9833 /* Store the partitioning step */
9834 comm->partition_step = step;
9836 /* Increase the DD partitioning counter */
9838 /* The state currently matches this DD partitioning count, store it */
9839 state_local->ddp_count = dd->ddp_count;
9842 /* The DD master node knows the complete cg distribution,
9843 * store the count so we can possibly skip the cg info communication.
9845 comm->master_cg_ddp_count = (bSortCG ? 0 : dd->ddp_count);
9848 if (comm->DD_debug > 0)
9850 /* Set the env var GMX_DD_DEBUG if you suspect corrupted indices */
9851 check_index_consistency(dd, top_global->natoms, ncg_mtop(top_global),
9852 "after partitioning");