src/mdlib/sim_util.c

   1 /* -*- mode: c; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4; c-file-style: "stroustrup"; -*-
   2  *
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  *          GROningen MAchine for Chemical Simulations
   9  *
  10  *                        VERSION 3.2.0
  11  * Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
  12  * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
  13  * Copyright (c) 2001-2004, The GROMACS development team,
  14  * check out http://www.gromacs.org for more information.
  15
  16  * This program is free software; you can redistribute it and/or
  17  * modify it under the terms of the GNU General Public License
  18  * as published by the Free Software Foundation; either version 2
  19  * of the License, or (at your option) any later version.
  20  *
  21  * If you want to redistribute modifications, please consider that
  22  * scientific software is very special. Version control is crucial -
  23  * bugs must be traceable. We will be happy to consider code for
  24  * inclusion in the official distribution, but derived work must not
  25  * be called official GROMACS. Details are found in the README & COPYING
  26  * files - if they are missing, get the official version at www.gromacs.org.
  27  *
  28  * To help us fund GROMACS development, we humbly ask that you cite
  29  * the papers on the package - you can find them in the top README file.
  30  *
  31  * For more info, check our website at http://www.gromacs.org
  32  *
  33  * And Hey:
  34  * GROwing Monsters And Cloning Shrimps
  35  */
  36 #ifdef HAVE_CONFIG_H
  37 #include <config.h>
  38 #endif
  39
  40 #ifdef GMX_CRAY_XT3
  41 #include<catamount/dclock.h>
  42 #endif
  43
  44
  45 #include <stdio.h>
  46 #include <time.h>
  47 #ifdef HAVE_SYS_TIME_H
  48 #include <sys/time.h>
  49 #endif
  50 #include <math.h>
  51 #include "typedefs.h"
  52 #include "string2.h"
  53 #include "gmxfio.h"
  54 #include "smalloc.h"
  55 #include "names.h"
  56 #include "confio.h"
  57 #include "mvdata.h"
  58 #include "txtdump.h"
  59 #include "pbc.h"
  60 #include "chargegroup.h"
  61 #include "vec.h"
  62 #include <time.h>
  63 #include "nrnb.h"
  64 #include "mshift.h"
  65 #include "mdrun.h"
  66 #include "sim_util.h"
  67 #include "update.h"
  68 #include "physics.h"
  69 #include "main.h"
  70 #include "mdatoms.h"
  71 #include "force.h"
  72 #include "bondf.h"
  73 #include "pme.h"
  74 #include "disre.h"
  75 #include "orires.h"
  76 #include "network.h"
  77 #include "calcmu.h"
  78 #include "constr.h"
  79 #include "xvgr.h"
  80 #include "trnio.h"
  81 #include "xtcio.h"
  82 #include "copyrite.h"
  83 #include "pull_rotation.h"
  84 #include "gmx_random.h"
  85 #include "mpelogging.h"
  86 #include "domdec.h"
  87 #include "partdec.h"
  88 #include "gmx_wallcycle.h"
  89 #include "genborn.h"
  90 #include "nbnxn_search.h"
  91 #include "nbnxn_kernels/nbnxn_kernel_ref.h"
  92 #include "nbnxn_kernels/nbnxn_kernel_x86_simd128.h"
  93 #include "nbnxn_kernels/nbnxn_kernel_x86_simd256.h"
  94 #include "nbnxn_kernels/nbnxn_kernel_gpu_ref.h"
  95
  96 #ifdef GMX_LIB_MPI
  97 #include <mpi.h>
  98 #endif
  99 #ifdef GMX_THREAD_MPI
 100 #include "tmpi.h"
 101 #endif
 102
 103 #include "adress.h"
 104 #include "qmmm.h"
 105
 106 #include "nbnxn_cuda_data_mgmt.h"
 107 #include "nbnxn_cuda/nbnxn_cuda.h"
 108
 109 #if 0
 110 typedef struct gmx_timeprint {
 111
 112 } t_gmx_timeprint;
 113 #endif
 114
 115 /* Portable version of ctime_r implemented in src/gmxlib/string2.c, but we do not want it declared in public installed headers */
 116 char *
 117 gmx_ctime_r(const time_t *clock,char *buf, int n);
 118
 119
 120 double
 121 gmx_gettime()
 122 {
 123 #ifdef HAVE_GETTIMEOFDAY
 124         struct timeval t;
 125         double seconds;
 126
 127         gettimeofday(&t,NULL);
 128
 129         seconds = (double) t.tv_sec + 1e-6*(double)t.tv_usec;
 130
 131         return seconds;
 132 #else
 133         double  seconds;
 134
 135         seconds = time(NULL);
 136
 137         return seconds;
 138 #endif
 139 }
 140
 141
 142 #define difftime(end,start) ((double)(end)-(double)(start))
 143
 144 void print_time(FILE *out,gmx_runtime_t *runtime,gmx_large_int_t step,
 145                 t_inputrec *ir, t_commrec *cr)
 146 {
 147     time_t finish;
 148     char   timebuf[STRLEN];
 149     double dt;
 150     char buf[48];
 151
 152 #ifndef GMX_THREAD_MPI
 153     if (!PAR(cr))
 154 #endif
 155     {
 156         fprintf(out,"\r");
 157     }
 158     fprintf(out,"step %s",gmx_step_str(step,buf));
 159     if ((step >= ir->nstlist))
 160     {
 161         runtime->last = gmx_gettime();
 162         dt = difftime(runtime->last,runtime->real);
 163         runtime->time_per_step = dt/(step - ir->init_step + 1);
 164
 165         dt = (ir->nsteps + ir->init_step - step)*runtime->time_per_step;
 166
 167         if (ir->nsteps >= 0)
 168         {
 169             if (dt >= 300)
 170             {
 171                 finish = (time_t) (runtime->last + dt);
 172                 gmx_ctime_r(&finish,timebuf,STRLEN);
 173                 sprintf(buf,"%s",timebuf);
 174                 buf[strlen(buf)-1]='\0';
 175                 fprintf(out,", will finish %s",buf);
 176             }
 177             else
 178                 fprintf(out,", remaining runtime: %5d s          ",(int)dt);
 179         }
 180         else
 181         {
 182             fprintf(out," performance: %.1f ns/day    ",
 183                     ir->delta_t/1000*24*60*60/runtime->time_per_step);
 184         }
 185     }
 186 #ifndef GMX_THREAD_MPI
 187     if (PAR(cr))
 188     {
 189         fprintf(out,"\n");
 190     }
 191 #endif
 192
 193     fflush(out);
 194 }
 195
 196 #ifdef NO_CLOCK
 197 #define clock() -1
 198 #endif
 199
 200 static double set_proctime(gmx_runtime_t *runtime)
 201 {
 202     double diff;
 203 #ifdef GMX_CRAY_XT3
 204     double prev;
 205
 206     prev = runtime->proc;
 207     runtime->proc = dclock();
 208
 209     diff = runtime->proc - prev;
 210 #else
 211     clock_t prev;
 212
 213     prev = runtime->proc;
 214     runtime->proc = clock();
 215
 216     diff = (double)(runtime->proc - prev)/(double)CLOCKS_PER_SEC;
 217 #endif
 218     if (diff < 0)
 219     {
 220         /* The counter has probably looped, ignore this data */
 221         diff = 0;
 222     }
 223
 224     return diff;
 225 }
 226
 227 void runtime_start(gmx_runtime_t *runtime)
 228 {
 229     runtime->real = gmx_gettime();
 230     runtime->proc          = 0;
 231     set_proctime(runtime);
 232     runtime->realtime      = 0;
 233     runtime->proctime      = 0;
 234     runtime->last          = 0;
 235     runtime->time_per_step = 0;
 236 }
 237
 238 void runtime_end(gmx_runtime_t *runtime)
 239 {
 240     double now;
 241
 242     now = gmx_gettime();
 243
 244     runtime->proctime += set_proctime(runtime);
 245     runtime->realtime  = now - runtime->real;
 246     runtime->real      = now;
 247 }
 248
 249 void runtime_upd_proc(gmx_runtime_t *runtime)
 250 {
 251     runtime->proctime += set_proctime(runtime);
 252 }
 253
 254 void print_date_and_time(FILE *fplog,int nodeid,const char *title,
 255                          const gmx_runtime_t *runtime)
 256 {
 257     int i;
 258     char timebuf[STRLEN];
 259     char time_string[STRLEN];
 260     time_t tmptime;
 261
 262     if (fplog)
 263     {
 264         if (runtime != NULL)
 265         {
 266             tmptime = (time_t) runtime->real;
 267             gmx_ctime_r(&tmptime,timebuf,STRLEN);
 268         }
 269         else
 270         {
 271             tmptime = (time_t) gmx_gettime();
 272             gmx_ctime_r(&tmptime,timebuf,STRLEN);
 273         }
 274         for(i=0; timebuf[i]>=' '; i++)
 275         {
 276             time_string[i]=timebuf[i];
 277         }
 278         time_string[i]='\0';
 279
 280         fprintf(fplog,"%s on node %d %s\n",title,nodeid,time_string);
 281     }
 282 }
 283
 284 static void sum_forces(int start,int end,rvec f[],rvec flr[])
 285 {
 286   int i;
 287
 288   if (gmx_debug_at) {
 289     pr_rvecs(debug,0,"fsr",f+start,end-start);
 290     pr_rvecs(debug,0,"flr",flr+start,end-start);
 291   }
 292   for(i=start; (i<end); i++)
 293     rvec_inc(f[i],flr[i]);
 294 }
 295
 296 /*
 297  * calc_f_el calculates forces due to an electric field.
 298  *
 299  * force is kJ mol^-1 nm^-1 = e * kJ mol^-1 nm^-1 / e
 300  *
 301  * Et[] contains the parameters for the time dependent
 302  * part of the field (not yet used).
 303  * Ex[] contains the parameters for
 304  * the spatial dependent part of the field. You can have cool periodic
 305  * fields in principle, but only a constant field is supported
 306  * now.
 307  * The function should return the energy due to the electric field
 308  * (if any) but for now returns 0.
 309  *
 310  * WARNING:
 311  * There can be problems with the virial.
 312  * Since the field is not self-consistent this is unavoidable.
 313  * For neutral molecules the virial is correct within this approximation.
 314  * For neutral systems with many charged molecules the error is small.
 315  * But for systems with a net charge or a few charged molecules
 316  * the error can be significant when the field is high.
 317  * Solution: implement a self-consitent electric field into PME.
 318  */
 319 static void calc_f_el(FILE *fp,int  start,int homenr,
 320                       real charge[],rvec x[],rvec f[],
 321                       t_cosines Ex[],t_cosines Et[],double t)
 322 {
 323     rvec Ext;
 324     real t0;
 325     int  i,m;
 326
 327     for(m=0; (m<DIM); m++)
 328     {
 329         if (Et[m].n > 0)
 330         {
 331             if (Et[m].n == 3)
 332             {
 333                 t0 = Et[m].a[1];
 334                 Ext[m] = cos(Et[m].a[0]*(t-t0))*exp(-sqr(t-t0)/(2.0*sqr(Et[m].a[2])));
 335             }
 336             else
 337             {
 338                 Ext[m] = cos(Et[m].a[0]*t);
 339             }
 340         }
 341         else
 342         {
 343             Ext[m] = 1.0;
 344         }
 345         if (Ex[m].n > 0)
 346         {
 347             /* Convert the field strength from V/nm to MD-units */
 348             Ext[m] *= Ex[m].a[0]*FIELDFAC;
 349             for(i=start; (i<start+homenr); i++)
 350                 f[i][m] += charge[i]*Ext[m];
 351         }
 352         else
 353         {
 354             Ext[m] = 0;
 355         }
 356     }
 357     if (fp != NULL)
 358     {
 359         fprintf(fp,"%10g  %10g  %10g  %10g #FIELD\n",t,
 360                 Ext[XX]/FIELDFAC,Ext[YY]/FIELDFAC,Ext[ZZ]/FIELDFAC);
 361     }
 362 }
 363
 364 static void calc_virial(FILE *fplog,int start,int homenr,rvec x[],rvec f[],
 365                         tensor vir_part,t_graph *graph,matrix box,
 366                         t_nrnb *nrnb,const t_forcerec *fr,int ePBC)
 367 {
 368   int i,j;
 369   tensor virtest;
 370
 371   /* The short-range virial from surrounding boxes */
 372   clear_mat(vir_part);
 373   calc_vir(fplog,SHIFTS,fr->shift_vec,fr->fshift,vir_part,ePBC==epbcSCREW,box);
 374   inc_nrnb(nrnb,eNR_VIRIAL,SHIFTS);
 375
 376   /* Calculate partial virial, for local atoms only, based on short range.
 377    * Total virial is computed in global_stat, called from do_md
 378    */
 379   f_calc_vir(fplog,start,start+homenr,x,f,vir_part,graph,box);
 380   inc_nrnb(nrnb,eNR_VIRIAL,homenr);
 381
 382   /* Add position restraint contribution */
 383   for(i=0; i<DIM; i++) {
 384     vir_part[i][i] += fr->vir_diag_posres[i];
 385   }
 386
 387   /* Add wall contribution */
 388   for(i=0; i<DIM; i++) {
 389     vir_part[i][ZZ] += fr->vir_wall_z[i];
 390   }
 391
 392   if (debug)
 393     pr_rvecs(debug,0,"vir_part",vir_part,DIM);
 394 }
 395
 396 static void posres_wrapper(FILE *fplog,
 397                            int flags,
 398                            gmx_bool bSepDVDL,
 399                            t_inputrec *ir,
 400                            t_nrnb *nrnb,
 401                            gmx_localtop_t *top,
 402                            matrix box,rvec x[],
 403                            rvec f[],
 404                            gmx_enerdata_t *enerd,
 405                            real *lambda,
 406                            t_forcerec *fr)
 407 {
 408     t_pbc pbc;
 409     real  v,dvdl;
 410     int   i;
 411
 412     /* Position restraints always require full pbc */
 413     set_pbc(&pbc,ir->ePBC,box);
 414     dvdl = 0;
 415     v = posres(top->idef.il[F_POSRES].nr,top->idef.il[F_POSRES].iatoms,
 416                top->idef.iparams_posres,
 417                (const rvec*)x,fr->f_novirsum,fr->vir_diag_posres,
 418                ir->ePBC==epbcNONE ? NULL : &pbc,
 419                lambda[efptRESTRAINT],&dvdl,
 420                fr->rc_scaling,fr->ePBC,fr->posres_com,fr->posres_comB);
 421     if (bSepDVDL)
 422     {
 423         fprintf(fplog,sepdvdlformat,
 424                 interaction_function[F_POSRES].longname,v,dvdl);
 425     }
 426     enerd->term[F_POSRES] += v;
 427     /* If just the force constant changes, the FEP term is linear,
 428      * but if k changes, it is not.
 429      */
 430     enerd->dvdl_nonlin[efptRESTRAINT] += dvdl;
 431     inc_nrnb(nrnb,eNR_POSRES,top->idef.il[F_POSRES].nr/2);
 432
 433     if ((ir->fepvals->n_lambda > 0) && (flags & GMX_FORCE_DHDL))
 434     {
 435         for(i=0; i<enerd->n_lambda; i++)
 436         {
 437             real dvdl_dum,lambda_dum;
 438
 439             lambda_dum = (i==0 ? lambda[efptRESTRAINT] : ir->fepvals->all_lambda[efptRESTRAINT][i-1]);
 440             v = posres(top->idef.il[F_POSRES].nr,top->idef.il[F_POSRES].iatoms,
 441                        top->idef.iparams_posres,
 442                        (const rvec*)x,NULL,NULL,
 443                        ir->ePBC==epbcNONE ? NULL : &pbc,lambda_dum,&dvdl,
 444                        fr->rc_scaling,fr->ePBC,fr->posres_com,fr->posres_comB);
 445             enerd->enerpart_lambda[i] += v;
 446         }
 447     }
 448 }
 449
 450 static void pull_potential_wrapper(FILE *fplog,
 451                                    gmx_bool bSepDVDL,
 452                                    t_commrec *cr,
 453                                    t_inputrec *ir,
 454                                    matrix box,rvec x[],
 455                                    rvec f[],
 456                                    tensor vir_force,
 457                                    t_mdatoms *mdatoms,
 458                                    gmx_enerdata_t *enerd,
 459                                    real *lambda,
 460                                    double t)
 461 {
 462     t_pbc  pbc;
 463     real   dvdl;
 464
 465     /* Calculate the center of mass forces, this requires communication,
 466      * which is why pull_potential is called close to other communication.
 467      * The virial contribution is calculated directly,
 468      * which is why we call pull_potential after calc_virial.
 469      */
 470     set_pbc(&pbc,ir->ePBC,box);
 471     dvdl = 0;
 472     enerd->term[F_COM_PULL] +=
 473         pull_potential(ir->ePull,ir->pull,mdatoms,&pbc,
 474                        cr,t,lambda[efptRESTRAINT],x,f,vir_force,&dvdl);
 475     if (bSepDVDL)
 476     {
 477         fprintf(fplog,sepdvdlformat,"Com pull",enerd->term[F_COM_PULL],dvdl);
 478     }
 479     enerd->dvdl_lin[efptRESTRAINT] += dvdl;
 480 }
 481
 482 static void pme_receive_force_ener(FILE *fplog,
 483                                    gmx_bool bSepDVDL,
 484                                    t_commrec *cr,
 485                                    gmx_wallcycle_t wcycle,
 486                                    gmx_enerdata_t *enerd,
 487                                    t_forcerec *fr)
 488 {
 489     real   e,v,dvdl;
 490     float  cycles_ppdpme,cycles_seppme;
 491
 492     cycles_ppdpme = wallcycle_stop(wcycle,ewcPPDURINGPME);
 493     dd_cycles_add(cr->dd,cycles_ppdpme,ddCyclPPduringPME);
 494
 495     /* In case of node-splitting, the PP nodes receive the long-range
 496      * forces, virial and energy from the PME nodes here.
 497      */
 498     wallcycle_start(wcycle,ewcPP_PMEWAITRECVF);
 499     dvdl = 0;
 500     gmx_pme_receive_f(cr,fr->f_novirsum,fr->vir_el_recip,&e,&dvdl,
 501                       &cycles_seppme);
 502     if (bSepDVDL)
 503     {
 504         fprintf(fplog,sepdvdlformat,"PME mesh",e,dvdl);
 505     }
 506     enerd->term[F_COUL_RECIP] += e;
 507     enerd->dvdl_lin[efptCOUL] += dvdl;
 508     if (wcycle)
 509     {
 510         dd_cycles_add(cr->dd,cycles_seppme,ddCyclPME);
 511     }
 512     wallcycle_stop(wcycle,ewcPP_PMEWAITRECVF);
 513 }
 514
 515 static void print_large_forces(FILE *fp,t_mdatoms *md,t_commrec *cr,
 516                                gmx_large_int_t step,real pforce,rvec *x,rvec *f)
 517 {
 518   int  i;
 519   real pf2,fn2;
 520   char buf[STEPSTRSIZE];
 521
 522   pf2 = sqr(pforce);
 523   for(i=md->start; i<md->start+md->homenr; i++) {
 524     fn2 = norm2(f[i]);
 525     /* We also catch NAN, if the compiler does not optimize this away. */
 526     if (fn2 >= pf2 || fn2 != fn2) {
 527       fprintf(fp,"step %s  atom %6d  x %8.3f %8.3f %8.3f  force %12.5e\n",
 528               gmx_step_str(step,buf),
 529               ddglatnr(cr->dd,i),x[i][XX],x[i][YY],x[i][ZZ],sqrt(fn2));
 530     }
 531   }
 532 }
 533
 534 static void post_process_forces(FILE *fplog,
 535                                 t_commrec *cr,
 536                                 gmx_large_int_t step,
 537                                 t_nrnb *nrnb,gmx_wallcycle_t wcycle,
 538                                 gmx_localtop_t *top,
 539                                 matrix box,rvec x[],
 540                                 rvec f[],
 541                                 tensor vir_force,
 542                                 t_mdatoms *mdatoms,
 543                                 t_graph *graph,
 544                                 t_forcerec *fr,gmx_vsite_t *vsite,
 545                                 int flags)
 546 {
 547     if (fr->bF_NoVirSum)
 548     {
 549         if (vsite)
 550         {
 551             /* Spread the mesh force on virtual sites to the other particles...
 552              * This is parallellized. MPI communication is performed
 553              * if the constructing atoms aren't local.
 554              */
 555             wallcycle_start(wcycle,ewcVSITESPREAD);
 556             spread_vsite_f(fplog,vsite,x,fr->f_novirsum,NULL,
 557                            (flags & GMX_FORCE_VIRIAL),fr->vir_el_recip,
 558                            nrnb,
 559                            &top->idef,fr->ePBC,fr->bMolPBC,graph,box,cr);
 560             wallcycle_stop(wcycle,ewcVSITESPREAD);
 561         }
 562         if (flags & GMX_FORCE_VIRIAL)
 563         {
 564             /* Now add the forces, this is local */
 565             if (fr->bDomDec)
 566             {
 567                 sum_forces(0,fr->f_novirsum_n,f,fr->f_novirsum);
 568             }
 569             else
 570             {
 571                 sum_forces(mdatoms->start,mdatoms->start+mdatoms->homenr,
 572                            f,fr->f_novirsum);
 573             }
 574             if (EEL_FULL(fr->eeltype))
 575             {
 576                 /* Add the mesh contribution to the virial */
 577                 m_add(vir_force,fr->vir_el_recip,vir_force);
 578             }
 579             if (debug)
 580             {
 581                 pr_rvecs(debug,0,"vir_force",vir_force,DIM);
 582             }
 583         }
 584     }
 585
 586     if (fr->print_force >= 0)
 587     {
 588         print_large_forces(stderr,mdatoms,cr,step,fr->print_force,x,f);
 589     }
 590 }
 591
 592 static void do_nb_verlet(t_forcerec *fr,
 593                          interaction_const_t *ic,
 594                          gmx_enerdata_t *enerd,
 595                          int flags, int ilocality,
 596                          int clearF,
 597                          t_nrnb *nrnb,
 598                          gmx_wallcycle_t wcycle)
 599 {
 600     int     nnbl, kernel_type, sh_e;
 601     char    *env;
 602     nonbonded_verlet_group_t  *nbvg;
 603
 604     if (!(flags & GMX_FORCE_NONBONDED))
 605     {
 606         /* skip non-bonded calculation */
 607         return;
 608     }
 609
 610     nbvg = &fr->nbv->grp[ilocality];
 611
 612     /* CUDA kernel launch overhead is already timed separately */
 613     if (fr->cutoff_scheme != ecutsVERLET)
 614     {
 615         gmx_incons("Invalid cut-off scheme passed!");
 616     }
 617
 618     if (nbvg->kernel_type != nbk8x8x8_CUDA)
 619     {
 620         wallcycle_sub_start(wcycle, ewcsNONBONDED);
 621     }
 622     switch (nbvg->kernel_type)
 623     {
 624         case nbk4x4_PlainC:
 625             nbnxn_kernel_ref(&nbvg->nbl_lists,
 626                              nbvg->nbat, ic,
 627                              fr->shift_vec,
 628                              flags,
 629                              clearF,
 630                              fr->fshift[0],
 631                              enerd->grpp.ener[egCOULSR],
 632                              fr->bBHAM ?
 633                              enerd->grpp.ener[egBHAMSR] :
 634                              enerd->grpp.ener[egLJSR]);
 635             break;
 636
 637         case nbk4xN_X86_SIMD128:
 638             nbnxn_kernel_x86_simd128(&nbvg->nbl_lists,
 639                                      nbvg->nbat, ic,
 640                                      fr->shift_vec,
 641                                      flags,
 642                                      clearF,
 643                                      fr->fshift[0],
 644                                      enerd->grpp.ener[egCOULSR],
 645                                      fr->bBHAM ?
 646                                      enerd->grpp.ener[egBHAMSR] :
 647                                      enerd->grpp.ener[egLJSR]);
 648             break;
 649         case nbk4xN_X86_SIMD256:
 650             nbnxn_kernel_x86_simd256(&nbvg->nbl_lists,
 651                                      nbvg->nbat, ic,
 652                                      fr->shift_vec,
 653                                      flags,
 654                                      clearF,
 655                                      fr->fshift[0],
 656                                      enerd->grpp.ener[egCOULSR],
 657                                      fr->bBHAM ?
 658                                      enerd->grpp.ener[egBHAMSR] :
 659                                      enerd->grpp.ener[egLJSR]);
 660             break;
 661
 662         case nbk8x8x8_CUDA:
 663             nbnxn_cuda_launch_kernel(fr->nbv->cu_nbv, nbvg->nbat, flags, ilocality);
 664             break;
 665
 666         case nbk8x8x8_PlainC:
 667             nbnxn_kernel_gpu_ref(nbvg->nbl_lists.nbl[0],
 668                                  nbvg->nbat, ic,
 669                                  fr->shift_vec,
 670                                  flags,
 671                                  clearF,
 672                                  nbvg->nbat->out[0].f,
 673                                  fr->fshift[0],
 674                                  enerd->grpp.ener[egCOULSR],
 675                                  fr->bBHAM ?
 676                                  enerd->grpp.ener[egBHAMSR] :
 677                                  enerd->grpp.ener[egLJSR]);
 678             break;
 679
 680         default:
 681             gmx_incons("Invalid nonbonded kernel type passed!");
 682
 683     }
 684     if (nbvg->kernel_type != nbk8x8x8_CUDA)
 685     {
 686         wallcycle_sub_stop(wcycle, ewcsNONBONDED);
 687     }
 688
 689     /* In eNR_??? the nbnxn F+E kernels are always the F kernel + 1 */
 690     sh_e = ((flags & GMX_FORCE_ENERGY) ? 1 : 0);
 691     inc_nrnb(nrnb,
 692              ((EEL_RF(ic->eeltype) || ic->eeltype == eelCUT) ?
 693               eNR_NBNXN_LJ_RF : eNR_NBNXN_LJ_TAB) + sh_e,
 694              nbvg->nbl_lists.natpair_ljq);
 695     inc_nrnb(nrnb,eNR_NBNXN_LJ+sh_e,nbvg->nbl_lists.natpair_lj);
 696     inc_nrnb(nrnb,
 697              ((EEL_RF(ic->eeltype) || ic->eeltype == eelCUT) ?
 698               eNR_NBNXN_RF : eNR_NBNXN_TAB)+sh_e,
 699              nbvg->nbl_lists.natpair_q);
 700 }
 701
 702 void do_force_cutsVERLET(FILE *fplog,t_commrec *cr,
 703               t_inputrec *inputrec,
 704               gmx_large_int_t step,t_nrnb *nrnb,gmx_wallcycle_t wcycle,
 705               gmx_localtop_t *top,
 706               gmx_mtop_t *mtop,
 707               gmx_groups_t *groups,
 708               matrix box,rvec x[],history_t *hist,
 709               rvec f[],
 710               tensor vir_force,
 711               t_mdatoms *mdatoms,
 712               gmx_enerdata_t *enerd,t_fcdata *fcd,
 713               real *lambda,t_graph *graph,
 714               t_forcerec *fr, interaction_const_t *ic,
 715               gmx_vsite_t *vsite,rvec mu_tot,
 716               double t,FILE *field,gmx_edsam_t ed,
 717               gmx_bool bBornRadii,
 718               int flags)
 719 {
 720     int     cg0,cg1,i,j;
 721     int     start,homenr;
 722     int     nb_kernel_type;
 723     double  mu[2*DIM];
 724     gmx_bool   bSepDVDL,bStateChanged,bNS,bFillGrid,bCalcCGCM,bBS;
 725     gmx_bool   bDoLongRange,bDoForces,bSepLRF,bUseGPU,bUseOrEmulGPU;
 726     gmx_bool   bDiffKernels=FALSE;
 727     matrix  boxs;
 728     rvec    vzero,box_diag;
 729     real    e,v,dvdl;
 730     float  cycles_pme,cycles_force;
 731     nonbonded_verlet_t *nbv;
 732
 733     cycles_force = 0;
 734     nbv = fr->nbv;
 735     nb_kernel_type = fr->nbv->grp[0].kernel_type;
 736
 737     start  = mdatoms->start;
 738     homenr = mdatoms->homenr;
 739
 740     bSepDVDL = (fr->bSepDVDL && do_per_step(step,inputrec->nstlog));
 741
 742     clear_mat(vir_force);
 743
 744     cg0 = 0;
 745     if (DOMAINDECOMP(cr))
 746     {
 747         cg1 = cr->dd->ncg_tot;
 748     }
 749     else
 750     {
 751         cg1 = top->cgs.nr;
 752     }
 753     if (fr->n_tpi > 0)
 754     {
 755         cg1--;
 756     }
 757
 758     bStateChanged = (flags & GMX_FORCE_STATECHANGED);
 759     bNS           = (flags & GMX_FORCE_NS) && (fr->bAllvsAll==FALSE);
 760     bFillGrid     = (bNS && bStateChanged);
 761     bCalcCGCM     = (bFillGrid && !DOMAINDECOMP(cr));
 762     bDoLongRange  = (fr->bTwinRange && bNS && (flags & GMX_FORCE_DOLR));
 763     bDoForces     = (flags & GMX_FORCE_FORCES);
 764     bSepLRF       = (bDoLongRange && bDoForces && (flags & GMX_FORCE_SEPLRF));
 765     bUseGPU       = fr->nbv->bUseGPU;
 766     bUseOrEmulGPU = bUseGPU || (nbv->grp[0].kernel_type == nbk8x8x8_PlainC);
 767
 768     if (bStateChanged)
 769     {
 770         update_forcerec(fplog,fr,box);
 771
 772         if (NEED_MUTOT(*inputrec))
 773         {
 774             /* Calculate total (local) dipole moment in a temporary common array.
 775              * This makes it possible to sum them over nodes faster.
 776              */
 777             calc_mu(start,homenr,
 778                     x,mdatoms->chargeA,mdatoms->chargeB,mdatoms->nChargePerturbed,
 779                     mu,mu+DIM);
 780         }
 781     }
 782
 783     if (fr->ePBC != epbcNONE) {
 784         /* Compute shift vectors every step,
 785          * because of pressure coupling or box deformation!
 786          */
 787         if ((flags & GMX_FORCE_DYNAMICBOX) && bStateChanged)
 788             calc_shifts(box,fr->shift_vec);
 789
 790         if (bCalcCGCM) {
 791             put_atoms_in_box_omp(fr->ePBC,box,homenr,x);
 792             inc_nrnb(nrnb,eNR_SHIFTX,homenr);
 793         }
 794         else if (EI_ENERGY_MINIMIZATION(inputrec->eI) && graph) {
 795             unshift_self(graph,box,x);
 796         }
 797     }
 798
 799     nbnxn_atomdata_copy_shiftvec(flags & GMX_FORCE_DYNAMICBOX,
 800                                   fr->shift_vec,nbv->grp[0].nbat);
 801
 802 #ifdef GMX_MPI
 803     if (!(cr->duty & DUTY_PME)) {
 804         /* Send particle coordinates to the pme nodes.
 805          * Since this is only implemented for domain decomposition
 806          * and domain decomposition does not use the graph,
 807          * we do not need to worry about shifting.
 808          */
 809
 810         wallcycle_start(wcycle,ewcPP_PMESENDX);
 811         GMX_MPE_LOG(ev_send_coordinates_start);
 812
 813         bBS = (inputrec->nwall == 2);
 814         if (bBS) {
 815             copy_mat(box,boxs);
 816             svmul(inputrec->wall_ewald_zfac,boxs[ZZ],boxs[ZZ]);
 817         }
 818
 819         gmx_pme_send_x(cr,bBS ? boxs : box,x,
 820                        mdatoms->nChargePerturbed,lambda[efptCOUL],
 821                        (flags & (GMX_FORCE_VIRIAL | GMX_FORCE_ENERGY)),step);
 822
 823         GMX_MPE_LOG(ev_send_coordinates_finish);
 824         wallcycle_stop(wcycle,ewcPP_PMESENDX);
 825     }
 826 #endif /* GMX_MPI */
 827
 828     /* do gridding for pair search */
 829     if (bNS)
 830     {
 831         if (graph && bStateChanged)
 832         {
 833             /* Calculate intramolecular shift vectors to make molecules whole */
 834             mk_mshift(fplog,graph,fr->ePBC,box,x);
 835         }
 836
 837         clear_rvec(vzero);
 838         box_diag[XX] = box[XX][XX];
 839         box_diag[YY] = box[YY][YY];
 840         box_diag[ZZ] = box[ZZ][ZZ];
 841
 842         wallcycle_start(wcycle,ewcNS);
 843         if (!fr->bDomDec)
 844         {
 845             wallcycle_sub_start(wcycle,ewcsNBS_GRID_LOCAL);
 846             nbnxn_put_on_grid(nbv->nbs,fr->ePBC,box,
 847                               0,vzero,box_diag,
 848                               0,mdatoms->homenr,-1,fr->cginfo,x,
 849                               0,NULL,
 850                               nbv->grp[eintLocal].kernel_type,
 851                               nbv->grp[eintLocal].nbat);
 852             wallcycle_sub_stop(wcycle,ewcsNBS_GRID_LOCAL);
 853         }
 854         else
 855         {
 856             wallcycle_sub_start(wcycle,ewcsNBS_GRID_NONLOCAL);
 857             nbnxn_put_on_grid_nonlocal(nbv->nbs,domdec_zones(cr->dd),
 858                                        fr->cginfo,x,
 859                                        nbv->grp[eintNonlocal].kernel_type,
 860                                        nbv->grp[eintNonlocal].nbat);
 861             wallcycle_sub_stop(wcycle,ewcsNBS_GRID_NONLOCAL);
 862         }
 863
 864         if (nbv->ngrp == 1 ||
 865             nbv->grp[eintNonlocal].nbat == nbv->grp[eintLocal].nbat)
 866         {
 867             nbnxn_atomdata_set(nbv->grp[eintLocal].nbat,eatAll,
 868                                 nbv->nbs,mdatoms,fr->cginfo);
 869         }
 870         else
 871         {
 872             nbnxn_atomdata_set(nbv->grp[eintLocal].nbat,eatLocal,
 873                                 nbv->nbs,mdatoms,fr->cginfo);
 874             nbnxn_atomdata_set(nbv->grp[eintNonlocal].nbat,eatAll,
 875                                 nbv->nbs,mdatoms,fr->cginfo);
 876         }
 877         wallcycle_stop(wcycle, ewcNS);
 878     }
 879
 880     /* initialize the GPU atom data and copy shift vector */
 881     if (bUseGPU)
 882     {
 883         if (bNS)
 884         {
 885             wallcycle_start_nocount(wcycle, ewcLAUNCH_GPU_NB);
 886             nbnxn_cuda_init_atomdata(nbv->cu_nbv, nbv->grp[eintLocal].nbat);
 887             wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB);
 888         }
 889
 890         wallcycle_start_nocount(wcycle, ewcLAUNCH_GPU_NB);
 891         nbnxn_cuda_upload_shiftvec(nbv->cu_nbv, nbv->grp[eintLocal].nbat);
 892         wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB);
 893     }
 894
 895     /* do local pair search */
 896     if (bNS)
 897     {
 898         wallcycle_start_nocount(wcycle,ewcNS);
 899         wallcycle_sub_start(wcycle,ewcsNBS_SEARCH_LOCAL);
 900         nbnxn_make_pairlist(nbv->nbs,nbv->grp[eintLocal].nbat,
 901                             &top->excls,
 902                             ic->rlist,
 903                             nbv->min_ci_balanced,
 904                             &nbv->grp[eintLocal].nbl_lists,
 905                             eintLocal,
 906                             nbv->grp[eintLocal].kernel_type,
 907                             nrnb);
 908         wallcycle_sub_stop(wcycle,ewcsNBS_SEARCH_LOCAL);
 909
 910         if (bUseGPU)
 911         {
 912             /* initialize local pair-list on the GPU */
 913             nbnxn_cuda_init_pairlist(nbv->cu_nbv,
 914                                      nbv->grp[eintLocal].nbl_lists.nbl[0],
 915                                      eintLocal);
 916         }
 917         wallcycle_stop(wcycle, ewcNS);
 918     }
 919     else
 920     {
 921         wallcycle_start(wcycle, ewcNB_XF_BUF_OPS);
 922         wallcycle_sub_start(wcycle, ewcsNB_X_BUF_OPS);
 923         nbnxn_atomdata_copy_x_to_nbat_x(nbv->nbs,eatLocal,FALSE,x,
 924                                         nbv->grp[eintLocal].nbat);
 925         wallcycle_sub_stop(wcycle, ewcsNB_X_BUF_OPS);
 926         wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS);
 927     }
 928
 929     if (bUseGPU)
 930     {
 931         wallcycle_start(wcycle,ewcLAUNCH_GPU_NB);
 932         /* launch local nonbonded F on GPU */
 933         do_nb_verlet(fr, ic, enerd, flags, eintLocal, enbvClearFNo,
 934                      nrnb, wcycle);
 935         wallcycle_stop(wcycle,ewcLAUNCH_GPU_NB);
 936     }
 937
 938     /* Communicate coordinates and sum dipole if necessary +
 939        do non-local pair search */
 940     if (DOMAINDECOMP(cr))
 941     {
 942         bDiffKernels = (nbv->grp[eintNonlocal].kernel_type !=
 943                         nbv->grp[eintLocal].kernel_type);
 944
 945         if (bDiffKernels)
 946         {
 947             /* With GPU+CPU non-bonded calculations we need to copy
 948              * the local coordinates to the non-local nbat struct
 949              * (in CPU format) as the non-local kernel call also
 950              * calculates the local - non-local interactions.
 951              */
 952             wallcycle_start(wcycle, ewcNB_XF_BUF_OPS);
 953             wallcycle_sub_start(wcycle, ewcsNB_X_BUF_OPS);
 954             nbnxn_atomdata_copy_x_to_nbat_x(nbv->nbs,eatLocal,TRUE,x,
 955                                              nbv->grp[eintNonlocal].nbat);
 956             wallcycle_sub_stop(wcycle, ewcsNB_X_BUF_OPS);
 957             wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS);
 958         }
 959
 960         if (bNS)
 961         {
 962             wallcycle_start_nocount(wcycle,ewcNS);
 963             wallcycle_sub_start(wcycle,ewcsNBS_SEARCH_NONLOCAL);
 964
 965             if (bDiffKernels)
 966             {
 967                 nbnxn_grid_add_simple(nbv->nbs,nbv->grp[eintNonlocal].nbat);
 968             }
 969
 970             nbnxn_make_pairlist(nbv->nbs,nbv->grp[eintNonlocal].nbat,
 971                                 &top->excls,
 972                                 ic->rlist,
 973                                 nbv->min_ci_balanced,
 974                                 &nbv->grp[eintNonlocal].nbl_lists,
 975                                 eintNonlocal,
 976                                 nbv->grp[eintNonlocal].kernel_type,
 977                                 nrnb);
 978
 979             wallcycle_sub_stop(wcycle,ewcsNBS_SEARCH_NONLOCAL);
 980
 981             if (nbv->grp[eintNonlocal].kernel_type == nbk8x8x8_CUDA)
 982             {
 983                 /* initialize non-local pair-list on the GPU */
 984                 nbnxn_cuda_init_pairlist(nbv->cu_nbv,
 985                                          nbv->grp[eintNonlocal].nbl_lists.nbl[0],
 986                                          eintNonlocal);
 987             }
 988             wallcycle_stop(wcycle,ewcNS);
 989         }
 990         else
 991         {
 992             wallcycle_start(wcycle,ewcMOVEX);
 993             dd_move_x(cr->dd,box,x);
 994
 995             /* When we don't need the total dipole we sum it in global_stat */
 996             if (bStateChanged && NEED_MUTOT(*inputrec))
 997             {
 998                 gmx_sumd(2*DIM,mu,cr);
 999             }
1000             wallcycle_stop(wcycle,ewcMOVEX);
1001
1002             wallcycle_start(wcycle, ewcNB_XF_BUF_OPS);
1003             wallcycle_sub_start(wcycle, ewcsNB_X_BUF_OPS);
1004             nbnxn_atomdata_copy_x_to_nbat_x(nbv->nbs,eatNonlocal,FALSE,x,
1005                                             nbv->grp[eintNonlocal].nbat);
1006             wallcycle_sub_stop(wcycle, ewcsNB_X_BUF_OPS);
1007             cycles_force += wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS);
1008         }
1009
1010         if (bUseGPU && !bDiffKernels)
1011         {
1012             wallcycle_start(wcycle,ewcLAUNCH_GPU_NB);
1013             /* launch non-local nonbonded F on GPU */
1014             do_nb_verlet(fr, ic, enerd, flags, eintNonlocal, enbvClearFNo,
1015                          nrnb, wcycle);
1016             cycles_force += wallcycle_stop(wcycle,ewcLAUNCH_GPU_NB);
1017         }
1018     }
1019
1020     if (bUseGPU)
1021     {
1022         /* launch D2H copy-back F */
1023         wallcycle_start_nocount(wcycle, ewcLAUNCH_GPU_NB);
1024         if (DOMAINDECOMP(cr) && !bDiffKernels)
1025         {
1026             nbnxn_cuda_launch_cpyback(nbv->cu_nbv, nbv->grp[eintNonlocal].nbat,
1027                                       flags, eatNonlocal);
1028         }
1029         nbnxn_cuda_launch_cpyback(nbv->cu_nbv, nbv->grp[eintLocal].nbat,
1030                                   flags, eatLocal);
1031         cycles_force += wallcycle_stop(wcycle,ewcLAUNCH_GPU_NB);
1032     }
1033
1034     if (bStateChanged && NEED_MUTOT(*inputrec))
1035     {
1036         if (PAR(cr))
1037         {
1038             gmx_sumd(2*DIM,mu,cr);
1039         }
1040
1041         for(i=0; i<2; i++)
1042         {
1043             for(j=0;j<DIM;j++)
1044             {
1045                 fr->mu_tot[i][j] = mu[i*DIM + j];
1046             }
1047         }
1048     }
1049     if (fr->efep == efepNO)
1050     {
1051         copy_rvec(fr->mu_tot[0],mu_tot);
1052     }
1053     else
1054     {
1055         for(j=0; j<DIM; j++)
1056         {
1057             mu_tot[j] =
1058                 (1.0 - lambda[efptCOUL])*fr->mu_tot[0][j] +
1059                 lambda[efptCOUL]*fr->mu_tot[1][j];
1060         }
1061     }
1062
1063     /* Reset energies */
1064     reset_enerdata(&(inputrec->opts),fr,bNS,enerd,MASTER(cr));
1065     clear_rvecs(SHIFTS,fr->fshift);
1066
1067     if (DOMAINDECOMP(cr))
1068     {
1069         if (!(cr->duty & DUTY_PME))
1070         {
1071             wallcycle_start(wcycle,ewcPPDURINGPME);
1072             dd_force_flop_start(cr->dd,nrnb);
1073         }
1074     }
1075
1076     /* Start the force cycle counter.
1077      * This counter is stopped in do_forcelow_level.
1078      * No parallel communication should occur while this counter is running,
1079      * since that will interfere with the dynamic load balancing.
1080      */
1081     wallcycle_start(wcycle,ewcFORCE);
1082     if (bDoForces)
1083     {
1084         /* Reset forces for which the virial is calculated separately:
1085          * PME/Ewald forces if necessary */
1086         if (fr->bF_NoVirSum)
1087         {
1088             if (flags & GMX_FORCE_VIRIAL)
1089             {
1090                 fr->f_novirsum = fr->f_novirsum_alloc;
1091                 GMX_BARRIER(cr->mpi_comm_mygroup);
1092                 if (fr->bDomDec)
1093                 {
1094                     clear_rvecs(fr->f_novirsum_n,fr->f_novirsum);
1095                 }
1096                 else
1097                 {
1098                     clear_rvecs(homenr,fr->f_novirsum+start);
1099                 }
1100                 GMX_BARRIER(cr->mpi_comm_mygroup);
1101             }
1102             else
1103             {
1104                 /* We are not calculating the pressure so we do not need
1105                  * a separate array for forces that do not contribute
1106                  * to the pressure.
1107                  */
1108                 fr->f_novirsum = f;
1109             }
1110         }
1111
1112         if (bSepLRF)
1113         {
1114             /* Add the long range forces to the short range forces */
1115             for(i=0; i<fr->natoms_force_constr; i++)
1116             {
1117                 copy_rvec(fr->f_twin[i],f[i]);
1118             }
1119         }
1120         else if (!(fr->bTwinRange && bNS))
1121         {
1122             /* Clear the short-range forces */
1123             clear_rvecs(fr->natoms_force_constr,f);
1124         }
1125
1126         clear_rvec(fr->vir_diag_posres);
1127
1128         GMX_BARRIER(cr->mpi_comm_mygroup);
1129     }
1130     if (inputrec->ePull == epullCONSTRAINT)
1131     {
1132         clear_pull_forces(inputrec->pull);
1133     }
1134
1135     /* update QMMMrec, if necessary */
1136     if(fr->bQMMM)
1137     {
1138         update_QMMMrec(cr,fr,x,mdatoms,box,top);
1139     }
1140
1141     if ((flags & GMX_FORCE_BONDED) && top->idef.il[F_POSRES].nr > 0)
1142     {
1143         posres_wrapper(fplog,flags,bSepDVDL,inputrec,nrnb,top,box,x,
1144                        f,enerd,lambda,fr);
1145     }
1146
1147     /* Compute the bonded and non-bonded energies and optionally forces */
1148     /* if we use the GPU turn off the nonbonded */
1149     do_force_lowlevel(fplog,step,fr,inputrec,&(top->idef),
1150                       cr,nrnb,wcycle,mdatoms,&(inputrec->opts),
1151                       x,hist,f,enerd,fcd,mtop,top,fr->born,
1152                       &(top->atomtypes),bBornRadii,box,
1153                       inputrec->fepvals,lambda,graph,&(top->excls),fr->mu_tot,
1154                       ((nb_kernel_type == nbk8x8x8_CUDA || nb_kernel_type == nbk8x8x8_PlainC)
1155                         ? flags&~GMX_FORCE_NONBONDED : flags),
1156                       &cycles_pme);
1157
1158     if (!bUseOrEmulGPU)
1159     {
1160         /* Maybe we should move this into do_force_lowlevel */
1161         do_nb_verlet(fr, ic, enerd, flags, eintLocal, enbvClearFYes,
1162                      nrnb, wcycle);
1163     }
1164
1165
1166     if (!bUseOrEmulGPU || bDiffKernels)
1167     {
1168         int aloc;
1169
1170         if (DOMAINDECOMP(cr))
1171         {
1172             do_nb_verlet(fr, ic, enerd, flags, eintNonlocal,
1173                          bDiffKernels ? enbvClearFYes : enbvClearFNo,
1174                          nrnb, wcycle);
1175         }
1176
1177         if (!bUseOrEmulGPU)
1178         {
1179             aloc = eintLocal;
1180         }
1181         else
1182         {
1183             aloc = eintNonlocal;
1184         }
1185
1186         /* Add all the non-bonded force to the normal force array.
1187          * This can be split into a local a non-local part when overlapping
1188          * communication with calculation with domain decomposition.
1189          */
1190         cycles_force += wallcycle_stop(wcycle,ewcFORCE);
1191         wallcycle_start(wcycle, ewcNB_XF_BUF_OPS);
1192         wallcycle_sub_start(wcycle, ewcsNB_F_BUF_OPS);
1193         nbnxn_atomdata_add_nbat_f_to_f(nbv->nbs,eatAll,nbv->grp[aloc].nbat,f);
1194         wallcycle_sub_stop(wcycle, ewcsNB_F_BUF_OPS);
1195         cycles_force += wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS);
1196         wallcycle_start_nocount(wcycle,ewcFORCE);
1197
1198         /* if there are multiple fshift output buffers reduce them */
1199         if ((flags & GMX_FORCE_VIRIAL) &&
1200             nbv->grp[aloc].nbl_lists.nnbl > 1)
1201         {
1202             nbnxn_atomdata_add_nbat_fshift_to_fshift(nbv->grp[aloc].nbat,
1203                                                       fr->fshift);
1204         }
1205     }
1206
1207     cycles_force += wallcycle_stop(wcycle,ewcFORCE);
1208     GMX_BARRIER(cr->mpi_comm_mygroup);
1209
1210     if (ed)
1211     {
1212         do_flood(fplog,cr,x,f,ed,box,step,bNS);
1213     }
1214
1215     if (bUseOrEmulGPU && !bDiffKernels)
1216     {
1217         /* wait for non-local forces (or calculate in emulation mode) */
1218         if (DOMAINDECOMP(cr))
1219         {
1220             if (bUseGPU)
1221             {
1222                 wallcycle_start(wcycle,ewcWAIT_GPU_NB_NL);
1223                 nbnxn_cuda_wait_gpu(nbv->cu_nbv,
1224                                     nbv->grp[eintNonlocal].nbat,
1225                                     flags, eatNonlocal,
1226                                     enerd->grpp.ener[egLJSR], enerd->grpp.ener[egCOULSR],
1227                                     fr->fshift);
1228                 cycles_force += wallcycle_stop(wcycle,ewcWAIT_GPU_NB_NL);
1229             }
1230             else
1231             {
1232                 wallcycle_start_nocount(wcycle,ewcFORCE);
1233                 do_nb_verlet(fr, ic, enerd, flags, eintNonlocal, enbvClearFYes,
1234                              nrnb, wcycle);
1235                 cycles_force += wallcycle_stop(wcycle,ewcFORCE);
1236             }
1237             wallcycle_start(wcycle, ewcNB_XF_BUF_OPS);
1238             wallcycle_sub_start(wcycle, ewcsNB_F_BUF_OPS);
1239             /* skip the reduction if there was no non-local work to do */
1240             if (nbv->grp[eintLocal].nbl_lists.nbl[0]->nsci > 0)
1241             {
1242                 nbnxn_atomdata_add_nbat_f_to_f(nbv->nbs,eatNonlocal,
1243                                                nbv->grp[eintNonlocal].nbat,f);
1244             }
1245             wallcycle_sub_stop(wcycle, ewcsNB_F_BUF_OPS);
1246             cycles_force += wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS);
1247         }
1248     }
1249
1250     if (bDoForces)
1251     {
1252         /* Communicate the forces */
1253         if (PAR(cr))
1254         {
1255             wallcycle_start(wcycle,ewcMOVEF);
1256             if (DOMAINDECOMP(cr))
1257             {
1258                 dd_move_f(cr->dd,f,fr->fshift);
1259                 /* Do we need to communicate the separate force array
1260                  * for terms that do not contribute to the single sum virial?
1261                  * Position restraints and electric fields do not introduce
1262                  * inter-cg forces, only full electrostatics methods do.
1263                  * When we do not calculate the virial, fr->f_novirsum = f,
1264                  * so we have already communicated these forces.
1265                  */
1266                 if (EEL_FULL(fr->eeltype) && cr->dd->n_intercg_excl &&
1267                     (flags & GMX_FORCE_VIRIAL))
1268                 {
1269                     dd_move_f(cr->dd,fr->f_novirsum,NULL);
1270                 }
1271                 if (bSepLRF)
1272                 {
1273                     /* We should not update the shift forces here,
1274                      * since f_twin is already included in f.
1275                      */
1276                     dd_move_f(cr->dd,fr->f_twin,NULL);
1277                 }
1278             }
1279             wallcycle_stop(wcycle,ewcMOVEF);
1280         }
1281     }
1282
1283     if (bUseOrEmulGPU)
1284     {
1285         /* wait for local forces (or calculate in emulation mode) */
1286         if (bUseGPU)
1287         {
1288             wallcycle_start(wcycle,ewcWAIT_GPU_NB_L);
1289             nbnxn_cuda_wait_gpu(nbv->cu_nbv,
1290                                 nbv->grp[eintLocal].nbat,
1291                                 flags, eatLocal,
1292                                 enerd->grpp.ener[egLJSR], enerd->grpp.ener[egCOULSR],
1293                                 fr->fshift);
1294             wallcycle_stop(wcycle,ewcWAIT_GPU_NB_L);
1295
1296             /* now clear the GPU outputs while we finish the step on the CPU */
1297             nbnxn_cuda_clear_outputs(nbv->cu_nbv, flags);
1298         }
1299         else
1300         {
1301             wallcycle_start_nocount(wcycle,ewcFORCE);
1302             do_nb_verlet(fr, ic, enerd, flags, eintLocal,
1303                          DOMAINDECOMP(cr) ? enbvClearFNo : enbvClearFYes,
1304                          nrnb, wcycle);
1305             wallcycle_stop(wcycle,ewcFORCE);
1306         }
1307         wallcycle_start(wcycle, ewcNB_XF_BUF_OPS);
1308         wallcycle_sub_start(wcycle, ewcsNB_F_BUF_OPS);
1309         if (nbv->grp[eintLocal].nbl_lists.nbl[0]->nsci > 0)
1310         {
1311             /* skip the reduction if there was no non-local work to do */
1312             nbnxn_atomdata_add_nbat_f_to_f(nbv->nbs,eatLocal,
1313                                            nbv->grp[eintLocal].nbat,f);
1314         }
1315         wallcycle_sub_stop(wcycle, ewcsNB_F_BUF_OPS);
1316         wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS);
1317     }
1318
1319     if (DOMAINDECOMP(cr))
1320     {
1321         dd_force_flop_stop(cr->dd,nrnb);
1322         if (wcycle)
1323         {
1324             dd_cycles_add(cr->dd,cycles_force-cycles_pme,ddCyclF);
1325         }
1326     }
1327
1328     if (bDoForces)
1329     {
1330         if (IR_ELEC_FIELD(*inputrec))
1331         {
1332             /* Compute forces due to electric field */
1333             calc_f_el(MASTER(cr) ? field : NULL,
1334                       start,homenr,mdatoms->chargeA,x,fr->f_novirsum,
1335                       inputrec->ex,inputrec->et,t);
1336         }
1337
1338         /* If we have NoVirSum forces, but we do not calculate the virial,
1339          * we sum fr->f_novirum=f later.
1340          */
1341         if (vsite && !(fr->bF_NoVirSum && !(flags & GMX_FORCE_VIRIAL)))
1342         {
1343             wallcycle_start(wcycle,ewcVSITESPREAD);
1344             spread_vsite_f(fplog,vsite,x,f,fr->fshift,FALSE,NULL,nrnb,
1345                            &top->idef,fr->ePBC,fr->bMolPBC,graph,box,cr);
1346             wallcycle_stop(wcycle,ewcVSITESPREAD);
1347
1348             if (bSepLRF)
1349             {
1350                 wallcycle_start(wcycle,ewcVSITESPREAD);
1351                 spread_vsite_f(fplog,vsite,x,fr->f_twin,NULL,FALSE,NULL,
1352                                nrnb,
1353                                &top->idef,fr->ePBC,fr->bMolPBC,graph,box,cr);
1354                 wallcycle_stop(wcycle,ewcVSITESPREAD);
1355             }
1356         }
1357
1358         if (flags & GMX_FORCE_VIRIAL)
1359         {
1360             /* Calculation of the virial must be done after vsites! */
1361             calc_virial(fplog,mdatoms->start,mdatoms->homenr,x,f,
1362                         vir_force,graph,box,nrnb,fr,inputrec->ePBC);
1363         }
1364     }
1365
1366     if (inputrec->ePull == epullUMBRELLA || inputrec->ePull == epullCONST_F)
1367     {
1368         pull_potential_wrapper(fplog,bSepDVDL,cr,inputrec,box,x,
1369                                f,vir_force,mdatoms,enerd,lambda,t);
1370     }
1371
1372     if (PAR(cr) && !(cr->duty & DUTY_PME))
1373     {
1374         /* In case of node-splitting, the PP nodes receive the long-range
1375          * forces, virial and energy from the PME nodes here.
1376          */
1377         pme_receive_force_ener(fplog,bSepDVDL,cr,wcycle,enerd,fr);
1378     }
1379
1380     if (bDoForces)
1381     {
1382         post_process_forces(fplog,cr,step,nrnb,wcycle,
1383                             top,box,x,f,vir_force,mdatoms,graph,fr,vsite,
1384                             flags);
1385     }
1386
1387     /* Sum the potential energy terms from group contributions */
1388     sum_epot(&(inputrec->opts),enerd);
1389 }
1390
1391 void do_force_cutsGROUP(FILE *fplog,t_commrec *cr,
1392               t_inputrec *inputrec,
1393               gmx_large_int_t step,t_nrnb *nrnb,gmx_wallcycle_t wcycle,
1394               gmx_localtop_t *top,
1395               gmx_mtop_t *mtop,
1396               gmx_groups_t *groups,
1397               matrix box,rvec x[],history_t *hist,
1398               rvec f[],
1399               tensor vir_force,
1400               t_mdatoms *mdatoms,
1401               gmx_enerdata_t *enerd,t_fcdata *fcd,
1402               real *lambda,t_graph *graph,
1403               t_forcerec *fr,gmx_vsite_t *vsite,rvec mu_tot,
1404               double t,FILE *field,gmx_edsam_t ed,
1405               gmx_bool bBornRadii,
1406               int flags)
1407 {
1408     int    cg0,cg1,i,j;
1409     int    start,homenr;
1410     double mu[2*DIM];
1411     gmx_bool   bSepDVDL,bStateChanged,bNS,bFillGrid,bCalcCGCM,bBS;
1412     gmx_bool   bDoLongRange,bDoForces,bSepLRF;
1413     gmx_bool   bDoAdressWF;
1414     matrix boxs;
1415     rvec   vzero,box_diag;
1416     real   e,v,dvdlambda[efptNR];
1417     t_pbc  pbc;
1418     float  cycles_pme,cycles_force;
1419
1420     start  = mdatoms->start;
1421     homenr = mdatoms->homenr;
1422
1423     bSepDVDL = (fr->bSepDVDL && do_per_step(step,inputrec->nstlog));
1424
1425     clear_mat(vir_force);
1426
1427     if (PARTDECOMP(cr))
1428     {
1429         pd_cg_range(cr,&cg0,&cg1);
1430     }
1431     else
1432     {
1433         cg0 = 0;
1434         if (DOMAINDECOMP(cr))
1435         {
1436             cg1 = cr->dd->ncg_tot;
1437         }
1438         else
1439         {
1440             cg1 = top->cgs.nr;
1441         }
1442         if (fr->n_tpi > 0)
1443         {
1444             cg1--;
1445         }
1446     }
1447
1448     bStateChanged = (flags & GMX_FORCE_STATECHANGED);
1449     bNS           = (flags & GMX_FORCE_NS) && (fr->bAllvsAll==FALSE);
1450     bFillGrid     = (bNS && bStateChanged);
1451     bCalcCGCM     = (bFillGrid && !DOMAINDECOMP(cr));
1452     bDoLongRange  = (fr->bTwinRange && bNS && (flags & GMX_FORCE_DOLR));
1453     bDoForces     = (flags & GMX_FORCE_FORCES);
1454     bSepLRF       = (bDoLongRange && bDoForces && (flags & GMX_FORCE_SEPLRF));
1455     /* should probably move this to the forcerec since it doesn't change */
1456     bDoAdressWF   = ((fr->adress_type!=eAdressOff));
1457
1458     if (bStateChanged)
1459     {
1460         update_forcerec(fplog,fr,box);
1461
1462         if (NEED_MUTOT(*inputrec))
1463         {
1464             /* Calculate total (local) dipole moment in a temporary common array.
1465              * This makes it possible to sum them over nodes faster.
1466              */
1467             calc_mu(start,homenr,
1468                     x,mdatoms->chargeA,mdatoms->chargeB,mdatoms->nChargePerturbed,
1469                     mu,mu+DIM);
1470         }
1471     }
1472
1473     if (fr->ePBC != epbcNONE) {
1474         /* Compute shift vectors every step,
1475          * because of pressure coupling or box deformation!
1476          */
1477         if ((flags & GMX_FORCE_DYNAMICBOX) && bStateChanged)
1478             calc_shifts(box,fr->shift_vec);
1479
1480         if (bCalcCGCM) {
1481             put_charge_groups_in_box(fplog,cg0,cg1,fr->ePBC,box,
1482                     &(top->cgs),x,fr->cg_cm);
1483             inc_nrnb(nrnb,eNR_CGCM,homenr);
1484             inc_nrnb(nrnb,eNR_RESETX,cg1-cg0);
1485         }
1486         else if (EI_ENERGY_MINIMIZATION(inputrec->eI) && graph) {
1487             unshift_self(graph,box,x);
1488         }
1489     }
1490     else if (bCalcCGCM) {
1491         calc_cgcm(fplog,cg0,cg1,&(top->cgs),x,fr->cg_cm);
1492         inc_nrnb(nrnb,eNR_CGCM,homenr);
1493     }
1494
1495     if (bCalcCGCM) {
1496         if (PAR(cr)) {
1497             move_cgcm(fplog,cr,fr->cg_cm);
1498         }
1499         if (gmx_debug_at)
1500             pr_rvecs(debug,0,"cgcm",fr->cg_cm,top->cgs.nr);
1501     }
1502
1503 #ifdef GMX_MPI
1504     if (!(cr->duty & DUTY_PME)) {
1505         /* Send particle coordinates to the pme nodes.
1506          * Since this is only implemented for domain decomposition
1507          * and domain decomposition does not use the graph,
1508          * we do not need to worry about shifting.
1509          */
1510
1511         wallcycle_start(wcycle,ewcPP_PMESENDX);
1512         GMX_MPE_LOG(ev_send_coordinates_start);
1513
1514         bBS = (inputrec->nwall == 2);
1515         if (bBS) {
1516             copy_mat(box,boxs);
1517             svmul(inputrec->wall_ewald_zfac,boxs[ZZ],boxs[ZZ]);
1518         }
1519
1520         gmx_pme_send_x(cr,bBS ? boxs : box,x,
1521                        mdatoms->nChargePerturbed,lambda[efptCOUL],
1522                        (flags & (GMX_FORCE_VIRIAL | GMX_FORCE_ENERGY)),step);
1523
1524         GMX_MPE_LOG(ev_send_coordinates_finish);
1525         wallcycle_stop(wcycle,ewcPP_PMESENDX);
1526     }
1527 #endif /* GMX_MPI */
1528
1529     /* Communicate coordinates and sum dipole if necessary */
1530     if (PAR(cr))
1531     {
1532         wallcycle_start(wcycle,ewcMOVEX);
1533         if (DOMAINDECOMP(cr))
1534         {
1535             dd_move_x(cr->dd,box,x);
1536         }
1537         else
1538         {
1539             move_x(fplog,cr,GMX_LEFT,GMX_RIGHT,x,nrnb);
1540         }
1541         wallcycle_stop(wcycle,ewcMOVEX);
1542     }
1543
1544     /* update adress weight beforehand */
1545     if(bStateChanged && bDoAdressWF)
1546     {
1547         /* need pbc for adress weight calculation with pbc_dx */
1548         set_pbc(&pbc,inputrec->ePBC,box);
1549         if(fr->adress_site == eAdressSITEcog)
1550         {
1551             update_adress_weights_cog(top->idef.iparams,top->idef.il,x,fr,mdatoms,
1552                                       inputrec->ePBC==epbcNONE ? NULL : &pbc);
1553         }
1554         else if (fr->adress_site == eAdressSITEcom)
1555         {
1556             update_adress_weights_com(fplog,cg0,cg1,&(top->cgs),x,fr,mdatoms,
1557                                       inputrec->ePBC==epbcNONE ? NULL : &pbc);
1558         }
1559         else if (fr->adress_site == eAdressSITEatomatom){
1560             update_adress_weights_atom_per_atom(cg0,cg1,&(top->cgs),x,fr,mdatoms,
1561                                                 inputrec->ePBC==epbcNONE ? NULL : &pbc);
1562         }
1563         else
1564         {
1565             update_adress_weights_atom(cg0,cg1,&(top->cgs),x,fr,mdatoms,
1566                                        inputrec->ePBC==epbcNONE ? NULL : &pbc);
1567         }
1568     }
1569
1570     if (NEED_MUTOT(*inputrec))
1571     {
1572
1573         if (bStateChanged)
1574         {
1575             if (PAR(cr))
1576             {
1577                 gmx_sumd(2*DIM,mu,cr);
1578             }
1579             for(i=0; i<2; i++)
1580             {
1581                 for(j=0;j<DIM;j++)
1582                 {
1583                     fr->mu_tot[i][j] = mu[i*DIM + j];
1584                 }
1585             }
1586         }
1587         if (fr->efep == efepNO)
1588         {
1589             copy_rvec(fr->mu_tot[0],mu_tot);
1590         }
1591         else
1592         {
1593             for(j=0; j<DIM; j++)
1594             {
1595                 mu_tot[j] =
1596                     (1.0 - lambda[efptCOUL])*fr->mu_tot[0][j] + lambda[efptCOUL]*fr->mu_tot[1][j];
1597             }
1598         }
1599     }
1600
1601     /* Reset energies */
1602     reset_enerdata(&(inputrec->opts),fr,bNS,enerd,MASTER(cr));
1603     clear_rvecs(SHIFTS,fr->fshift);
1604
1605     if (bNS)
1606     {
1607         wallcycle_start(wcycle,ewcNS);
1608
1609         if (graph && bStateChanged)
1610         {
1611             /* Calculate intramolecular shift vectors to make molecules whole */
1612             mk_mshift(fplog,graph,fr->ePBC,box,x);
1613         }
1614
1615         /* Reset long range forces if necessary */
1616         if (fr->bTwinRange)
1617         {
1618             /* Reset the (long-range) forces if necessary */
1619             clear_rvecs(fr->natoms_force_constr,bSepLRF ? fr->f_twin : f);
1620         }
1621
1622         /* Do the actual neighbour searching and if twin range electrostatics
1623          * also do the calculation of long range forces and energies.
1624          */
1625         for (i=0;i<efptNR;i++) {dvdlambda[i] = 0;}
1626         ns(fplog,fr,x,box,
1627            groups,&(inputrec->opts),top,mdatoms,
1628            cr,nrnb,lambda,dvdlambda,&enerd->grpp,bFillGrid,
1629            bDoLongRange,bDoForces,bSepLRF ? fr->f_twin : f);
1630         if (bSepDVDL)
1631         {
1632             fprintf(fplog,sepdvdlformat,"LR non-bonded",0.0,dvdlambda);
1633         }
1634         enerd->dvdl_lin[efptVDW] += dvdlambda[efptVDW];
1635         enerd->dvdl_lin[efptCOUL] += dvdlambda[efptCOUL];
1636
1637         wallcycle_stop(wcycle,ewcNS);
1638     }
1639
1640     if (inputrec->implicit_solvent && bNS)
1641     {
1642         make_gb_nblist(cr,inputrec->gb_algorithm,inputrec->rlist,
1643                        x,box,fr,&top->idef,graph,fr->born);
1644     }
1645
1646     if (DOMAINDECOMP(cr))
1647     {
1648         if (!(cr->duty & DUTY_PME))
1649         {
1650             wallcycle_start(wcycle,ewcPPDURINGPME);
1651             dd_force_flop_start(cr->dd,nrnb);
1652         }
1653     }
1654
1655     if (inputrec->bRot)
1656     {
1657         /* Enforced rotation has its own cycle counter that starts after the collective
1658          * coordinates have been communicated. It is added to ddCyclF to allow
1659          * for proper load-balancing */
1660         wallcycle_start(wcycle,ewcROT);
1661         do_rotation(cr,inputrec,box,x,t,step,wcycle,bNS);
1662         wallcycle_stop(wcycle,ewcROT);
1663     }
1664
1665     /* Start the force cycle counter.
1666      * This counter is stopped in do_forcelow_level.
1667      * No parallel communication should occur while this counter is running,
1668      * since that will interfere with the dynamic load balancing.
1669      */
1670     wallcycle_start(wcycle,ewcFORCE);
1671
1672     if (bDoForces)
1673     {
1674         /* Reset forces for which the virial is calculated separately:
1675          * PME/Ewald forces if necessary */
1676         if (fr->bF_NoVirSum)
1677         {
1678             if (flags & GMX_FORCE_VIRIAL)
1679             {
1680                 fr->f_novirsum = fr->f_novirsum_alloc;
1681                 GMX_BARRIER(cr->mpi_comm_mygroup);
1682                 if (fr->bDomDec)
1683                 {
1684                     clear_rvecs(fr->f_novirsum_n,fr->f_novirsum);
1685                 }
1686                 else
1687                 {
1688                     clear_rvecs(homenr,fr->f_novirsum+start);
1689                 }
1690                 GMX_BARRIER(cr->mpi_comm_mygroup);
1691             }
1692             else
1693             {
1694                 /* We are not calculating the pressure so we do not need
1695                  * a separate array for forces that do not contribute
1696                  * to the pressure.
1697                  */
1698                 fr->f_novirsum = f;
1699             }
1700         }
1701
1702         if (bSepLRF)
1703         {
1704             /* Add the long range forces to the short range forces */
1705             for(i=0; i<fr->natoms_force_constr; i++)
1706             {
1707                 copy_rvec(fr->f_twin[i],f[i]);
1708             }
1709         }
1710         else if (!(fr->bTwinRange && bNS))
1711         {
1712             /* Clear the short-range forces */
1713             clear_rvecs(fr->natoms_force_constr,f);
1714         }
1715
1716         clear_rvec(fr->vir_diag_posres);
1717
1718         GMX_BARRIER(cr->mpi_comm_mygroup);
1719     }
1720     if (inputrec->ePull == epullCONSTRAINT)
1721     {
1722         clear_pull_forces(inputrec->pull);
1723     }
1724
1725     /* update QMMMrec, if necessary */
1726     if(fr->bQMMM)
1727     {
1728         update_QMMMrec(cr,fr,x,mdatoms,box,top);
1729     }
1730
1731     if ((flags & GMX_FORCE_BONDED) && top->idef.il[F_POSRES].nr > 0)
1732     {
1733         posres_wrapper(fplog,flags,bSepDVDL,inputrec,nrnb,top,box,x,
1734                        f,enerd,lambda,fr);
1735     }
1736
1737     /* Compute the bonded and non-bonded energies and optionally forces */
1738     do_force_lowlevel(fplog,step,fr,inputrec,&(top->idef),
1739                       cr,nrnb,wcycle,mdatoms,&(inputrec->opts),
1740                       x,hist,f,enerd,fcd,mtop,top,fr->born,
1741                       &(top->atomtypes),bBornRadii,box,
1742                       inputrec->fepvals,lambda,
1743                       graph,&(top->excls),fr->mu_tot,
1744                       flags,
1745                       &cycles_pme);
1746
1747     cycles_force = wallcycle_stop(wcycle,ewcFORCE);
1748     GMX_BARRIER(cr->mpi_comm_mygroup);
1749
1750     if (ed)
1751     {
1752         do_flood(fplog,cr,x,f,ed,box,step,bNS);
1753     }
1754
1755     if (DOMAINDECOMP(cr))
1756     {
1757         dd_force_flop_stop(cr->dd,nrnb);
1758         if (wcycle)
1759         {
1760             dd_cycles_add(cr->dd,cycles_force-cycles_pme,ddCyclF);
1761         }
1762     }
1763
1764     if (bDoForces)
1765     {
1766         if (IR_ELEC_FIELD(*inputrec))
1767         {
1768             /* Compute forces due to electric field */
1769             calc_f_el(MASTER(cr) ? field : NULL,
1770                       start,homenr,mdatoms->chargeA,x,fr->f_novirsum,
1771                       inputrec->ex,inputrec->et,t);
1772         }
1773
1774         if (bDoAdressWF && fr->adress_icor == eAdressICThermoForce)
1775         {
1776             /* Compute thermodynamic force in hybrid AdResS region */
1777             adress_thermo_force(start,homenr,&(top->cgs),x,fr->f_novirsum,fr,mdatoms,
1778                                 inputrec->ePBC==epbcNONE ? NULL : &pbc);
1779         }
1780
1781         /* Communicate the forces */
1782         if (PAR(cr))
1783         {
1784             wallcycle_start(wcycle,ewcMOVEF);
1785             if (DOMAINDECOMP(cr))
1786             {
1787                 dd_move_f(cr->dd,f,fr->fshift);
1788                 /* Do we need to communicate the separate force array
1789                  * for terms that do not contribute to the single sum virial?
1790                  * Position restraints and electric fields do not introduce
1791                  * inter-cg forces, only full electrostatics methods do.
1792                  * When we do not calculate the virial, fr->f_novirsum = f,
1793                  * so we have already communicated these forces.
1794                  */
1795                 if (EEL_FULL(fr->eeltype) && cr->dd->n_intercg_excl &&
1796                     (flags & GMX_FORCE_VIRIAL))
1797                 {
1798                     dd_move_f(cr->dd,fr->f_novirsum,NULL);
1799                 }
1800                 if (bSepLRF)
1801                 {
1802                     /* We should not update the shift forces here,
1803                      * since f_twin is already included in f.
1804                      */
1805                     dd_move_f(cr->dd,fr->f_twin,NULL);
1806                 }
1807             }
1808             else
1809             {
1810                 pd_move_f(cr,f,nrnb);
1811                 if (bSepLRF)
1812                 {
1813                     pd_move_f(cr,fr->f_twin,nrnb);
1814                 }
1815             }
1816             wallcycle_stop(wcycle,ewcMOVEF);
1817         }
1818
1819         /* If we have NoVirSum forces, but we do not calculate the virial,
1820          * we sum fr->f_novirum=f later.
1821          */
1822         if (vsite && !(fr->bF_NoVirSum && !(flags & GMX_FORCE_VIRIAL)))
1823         {
1824             wallcycle_start(wcycle,ewcVSITESPREAD);
1825             spread_vsite_f(fplog,vsite,x,f,fr->fshift,FALSE,NULL,nrnb,
1826                            &top->idef,fr->ePBC,fr->bMolPBC,graph,box,cr);
1827             wallcycle_stop(wcycle,ewcVSITESPREAD);
1828
1829             if (bSepLRF)
1830             {
1831                 wallcycle_start(wcycle,ewcVSITESPREAD);
1832                 spread_vsite_f(fplog,vsite,x,fr->f_twin,NULL,FALSE,NULL,
1833                                nrnb,
1834                                &top->idef,fr->ePBC,fr->bMolPBC,graph,box,cr);
1835                 wallcycle_stop(wcycle,ewcVSITESPREAD);
1836             }
1837         }
1838
1839         if (flags & GMX_FORCE_VIRIAL)
1840         {
1841             /* Calculation of the virial must be done after vsites! */
1842             calc_virial(fplog,mdatoms->start,mdatoms->homenr,x,f,
1843                         vir_force,graph,box,nrnb,fr,inputrec->ePBC);
1844         }
1845     }
1846
1847     if (inputrec->ePull == epullUMBRELLA || inputrec->ePull == epullCONST_F)
1848     {
1849         pull_potential_wrapper(fplog,bSepDVDL,cr,inputrec,box,x,
1850                                f,vir_force,mdatoms,enerd,lambda,t);
1851     }
1852
1853     /* Add the forces from enforced rotation potentials (if any) */
1854     if (inputrec->bRot)
1855     {
1856         wallcycle_start(wcycle,ewcROTadd);
1857         enerd->term[F_COM_PULL] += add_rot_forces(inputrec->rot, f, cr,step,t);
1858         wallcycle_stop(wcycle,ewcROTadd);
1859     }
1860
1861     if (PAR(cr) && !(cr->duty & DUTY_PME))
1862     {
1863         /* In case of node-splitting, the PP nodes receive the long-range
1864          * forces, virial and energy from the PME nodes here.
1865          */
1866         pme_receive_force_ener(fplog,bSepDVDL,cr,wcycle,enerd,fr);
1867     }
1868
1869     if (bDoForces)
1870     {
1871         post_process_forces(fplog,cr,step,nrnb,wcycle,
1872                             top,box,x,f,vir_force,mdatoms,graph,fr,vsite,
1873                             flags);
1874     }
1875
1876     /* Sum the potential energy terms from group contributions */
1877     sum_epot(&(inputrec->opts),enerd);
1878 }
1879
1880 void do_force(FILE *fplog,t_commrec *cr,
1881               t_inputrec *inputrec,
1882               gmx_large_int_t step,t_nrnb *nrnb,gmx_wallcycle_t wcycle,
1883               gmx_localtop_t *top,
1884               gmx_mtop_t *mtop,
1885               gmx_groups_t *groups,
1886               matrix box,rvec x[],history_t *hist,
1887               rvec f[],
1888               tensor vir_force,
1889               t_mdatoms *mdatoms,
1890               gmx_enerdata_t *enerd,t_fcdata *fcd,
1891               real *lambda,t_graph *graph,
1892               t_forcerec *fr,
1893               gmx_vsite_t *vsite,rvec mu_tot,
1894               double t,FILE *field,gmx_edsam_t ed,
1895               gmx_bool bBornRadii,
1896               int flags)
1897 {
1898     /* modify force flag if not doing nonbonded */
1899     if (!fr->bNonbonded)
1900     {
1901         flags &= ~GMX_FORCE_NONBONDED;
1902     }
1903
1904     switch (inputrec->cutoff_scheme)
1905     {
1906         case ecutsVERLET:
1907             do_force_cutsVERLET(fplog, cr, inputrec,
1908                                 step, nrnb, wcycle,
1909                                 top, mtop,
1910                                 groups,
1911                                 box, x, hist,
1912                                 f, vir_force,
1913                                 mdatoms,
1914                                 enerd, fcd,
1915                                 lambda, graph,
1916                                 fr, fr->ic,
1917                                 vsite, mu_tot,
1918                                 t, field, ed,
1919                                 bBornRadii,
1920                                 flags);
1921             break;
1922         case ecutsGROUP:
1923              do_force_cutsGROUP(fplog, cr, inputrec,
1924                                 step, nrnb, wcycle,
1925                                 top, mtop,
1926                                 groups,
1927                                 box, x, hist,
1928                                 f, vir_force,
1929                                 mdatoms,
1930                                 enerd, fcd,
1931                                 lambda, graph,
1932                                 fr, vsite, mu_tot,
1933                                 t, field, ed,
1934                                 bBornRadii,
1935                                 flags);
1936             break;
1937         default:
1938             gmx_incons("Invalid cut-off scheme passed!");
1939     }
1940 }
1941
1942
1943 void do_constrain_first(FILE *fplog,gmx_constr_t constr,
1944                         t_inputrec *ir,t_mdatoms *md,
1945                         t_state *state,rvec *f,
1946                         t_graph *graph,t_commrec *cr,t_nrnb *nrnb,
1947                         t_forcerec *fr, gmx_localtop_t *top, tensor shake_vir)
1948 {
1949     int    i,m,start,end;
1950     gmx_large_int_t step;
1951     real   dt=ir->delta_t;
1952     real   dvdl_dum;
1953     rvec   *savex;
1954
1955     snew(savex,state->natoms);
1956
1957     start = md->start;
1958     end   = md->homenr + start;
1959
1960     if (debug)
1961         fprintf(debug,"vcm: start=%d, homenr=%d, end=%d\n",
1962                 start,md->homenr,end);
1963     /* Do a first constrain to reset particles... */
1964     step = ir->init_step;
1965     if (fplog)
1966     {
1967         char buf[STEPSTRSIZE];
1968         fprintf(fplog,"\nConstraining the starting coordinates (step %s)\n",
1969                 gmx_step_str(step,buf));
1970     }
1971     dvdl_dum = 0;
1972
1973     /* constrain the current position */
1974     constrain(NULL,TRUE,FALSE,constr,&(top->idef),
1975               ir,NULL,cr,step,0,md,
1976               state->x,state->x,NULL,
1977               fr->bMolPBC,state->box,
1978               state->lambda[efptBONDED],&dvdl_dum,
1979               NULL,NULL,nrnb,econqCoord,
1980               ir->epc==epcMTTK,state->veta,state->veta);
1981     if (EI_VV(ir->eI))
1982     {
1983         /* constrain the inital velocity, and save it */
1984         /* also may be useful if we need the ekin from the halfstep for velocity verlet */
1985         /* might not yet treat veta correctly */
1986         constrain(NULL,TRUE,FALSE,constr,&(top->idef),
1987                   ir,NULL,cr,step,0,md,
1988                   state->x,state->v,state->v,
1989                   fr->bMolPBC,state->box,
1990                   state->lambda[efptBONDED],&dvdl_dum,
1991                   NULL,NULL,nrnb,econqVeloc,
1992                   ir->epc==epcMTTK,state->veta,state->veta);
1993     }
1994     /* constrain the inital velocities at t-dt/2 */
1995     if (EI_STATE_VELOCITY(ir->eI) && ir->eI!=eiVV)
1996     {
1997         for(i=start; (i<end); i++)
1998         {
1999             for(m=0; (m<DIM); m++)
2000             {
2001                 /* Reverse the velocity */
2002                 state->v[i][m] = -state->v[i][m];
2003                 /* Store the position at t-dt in buf */
2004                 savex[i][m] = state->x[i][m] + dt*state->v[i][m];
2005             }
2006         }
2007     /* Shake the positions at t=-dt with the positions at t=0
2008      * as reference coordinates.
2009          */
2010         if (fplog)
2011         {
2012             char buf[STEPSTRSIZE];
2013             fprintf(fplog,"\nConstraining the coordinates at t0-dt (step %s)\n",
2014                     gmx_step_str(step,buf));
2015         }
2016         dvdl_dum = 0;
2017         constrain(NULL,TRUE,FALSE,constr,&(top->idef),
2018                   ir,NULL,cr,step,-1,md,
2019                   state->x,savex,NULL,
2020                   fr->bMolPBC,state->box,
2021                   state->lambda[efptBONDED],&dvdl_dum,
2022                   state->v,NULL,nrnb,econqCoord,
2023                   ir->epc==epcMTTK,state->veta,state->veta);
2024
2025         for(i=start; i<end; i++) {
2026             for(m=0; m<DIM; m++) {
2027                 /* Re-reverse the velocities */
2028                 state->v[i][m] = -state->v[i][m];
2029             }
2030         }
2031     }
2032     sfree(savex);
2033 }
2034
2035 void calc_enervirdiff(FILE *fplog,int eDispCorr,t_forcerec *fr)
2036 {
2037   double eners[2],virs[2],enersum,virsum,y0,f,g,h;
2038   double r0,r1,r,rc3,rc9,ea,eb,ec,pa,pb,pc,pd;
2039   double invscale,invscale2,invscale3;
2040   int    ri0,ri1,ri,i,offstart,offset;
2041   real   scale,*vdwtab;
2042
2043   fr->enershiftsix = 0;
2044   fr->enershifttwelve = 0;
2045   fr->enerdiffsix = 0;
2046   fr->enerdifftwelve = 0;
2047   fr->virdiffsix = 0;
2048   fr->virdifftwelve = 0;
2049
2050   if (eDispCorr != edispcNO) {
2051     for(i=0; i<2; i++) {
2052       eners[i] = 0;
2053       virs[i]  = 0;
2054     }
2055     if ((fr->vdwtype == evdwSWITCH) || (fr->vdwtype == evdwSHIFT)) {
2056       if (fr->rvdw_switch == 0)
2057         gmx_fatal(FARGS,
2058                   "With dispersion correction rvdw-switch can not be zero "
2059                   "for vdw-type = %s",evdw_names[fr->vdwtype]);
2060
2061       scale  = fr->nblists[0].tab.scale;
2062       vdwtab = fr->nblists[0].vdwtab;
2063
2064       /* Round the cut-offs to exact table values for precision */
2065       ri0 = floor(fr->rvdw_switch*scale);
2066       ri1 = ceil(fr->rvdw*scale);
2067       r0  = ri0/scale;
2068       r1  = ri1/scale;
2069       rc3 = r0*r0*r0;
2070       rc9  = rc3*rc3*rc3;
2071
2072       if (fr->vdwtype == evdwSHIFT) {
2073         /* Determine the constant energy shift below rvdw_switch */
2074         fr->enershiftsix    = (real)(-1.0/(rc3*rc3)) - vdwtab[8*ri0];
2075         fr->enershifttwelve = (real)( 1.0/(rc9*rc3)) - vdwtab[8*ri0 + 4];
2076       }
2077       /* Add the constant part from 0 to rvdw_switch.
2078        * This integration from 0 to rvdw_switch overcounts the number
2079        * of interactions by 1, as it also counts the self interaction.
2080        * We will correct for this later.
2081        */
2082       eners[0] += 4.0*M_PI*fr->enershiftsix*rc3/3.0;
2083       eners[1] += 4.0*M_PI*fr->enershifttwelve*rc3/3.0;
2084
2085       invscale = 1.0/(scale);
2086       invscale2 = invscale*invscale;
2087       invscale3 = invscale*invscale2;
2088
2089       /* following summation derived from cubic spline definition,
2090         Numerical Recipies in C, second edition, p. 113-116.  Exact
2091         for the cubic spline.  We first calculate the negative of
2092         the energy from rvdw to rvdw_switch, assuming that g(r)=1,
2093         and then add the more standard, abrupt cutoff correction to
2094         that result, yielding the long-range correction for a
2095         switched function.  We perform both the pressure and energy
2096         loops at the same time for simplicity, as the computational
2097         cost is low. */
2098
2099       for (i=0;i<2;i++) {
2100         enersum = 0.0; virsum = 0.0;
2101         if (i==0)
2102           offstart = 0;
2103         else
2104           offstart = 4;
2105         for (ri=ri0; ri<ri1; ri++) {
2106           r = ri*invscale;
2107           ea = invscale3;
2108           eb = 2.0*invscale2*r;
2109           ec = invscale*r*r;
2110
2111           pa = invscale3;
2112           pb = 3.0*invscale2*r;
2113           pc = 3.0*invscale*r*r;
2114           pd = r*r*r;
2115
2116           /* this "8" is from the packing in the vdwtab array - perhaps
2117             should be #define'ed? */
2118           offset = 8*ri + offstart;
2119           y0 = vdwtab[offset];
2120           f = vdwtab[offset+1];
2121           g = vdwtab[offset+2];
2122           h = vdwtab[offset+3];
2123
2124           enersum += y0*(ea/3 + eb/2 + ec) + f*(ea/4 + eb/3 + ec/2)+
2125             g*(ea/5 + eb/4 + ec/3) + h*(ea/6 + eb/5 + ec/4);
2126           virsum  +=  f*(pa/4 + pb/3 + pc/2 + pd) +
2127             2*g*(pa/5 + pb/4 + pc/3 + pd/2) + 3*h*(pa/6 + pb/5 + pc/4 + pd/3);
2128
2129         }
2130         enersum *= 4.0*M_PI;
2131         virsum  *= 4.0*M_PI;
2132         eners[i] -= enersum;
2133         virs[i]  -= virsum;
2134       }
2135
2136       /* now add the correction for rvdw_switch to infinity */
2137       eners[0] += -4.0*M_PI/(3.0*rc3);
2138       eners[1] +=  4.0*M_PI/(9.0*rc9);
2139       virs[0]  +=  8.0*M_PI/rc3;
2140       virs[1]  += -16.0*M_PI/(3.0*rc9);
2141     }
2142     else if ((fr->vdwtype == evdwCUT) || (fr->vdwtype == evdwUSER)) {
2143       if (fr->vdwtype == evdwUSER && fplog)
2144         fprintf(fplog,
2145                 "WARNING: using dispersion correction with user tables\n");
2146       rc3  = fr->rvdw*fr->rvdw*fr->rvdw;
2147       rc9  = rc3*rc3*rc3;
2148       /* Contribution beyond the cut-off */
2149       eners[0] += -4.0*M_PI/(3.0*rc3);
2150       eners[1] +=  4.0*M_PI/(9.0*rc9);
2151       if (fr->cutoff_scheme == ecutsVERLET && fr->ic->sh_invrc6 != 0) {
2152           /* Contribution within the cut-off */
2153           eners[0] += -4.0*M_PI/(3.0*rc3);
2154           eners[1] +=  4.0*M_PI/(3.0*rc9);
2155       }
2156       /* Contribution beyond the cut-off */
2157       virs[0]  +=  8.0*M_PI/rc3;
2158       virs[1]  += -16.0*M_PI/(3.0*rc9);
2159     } else {
2160       gmx_fatal(FARGS,
2161                 "Dispersion correction is not implemented for vdw-type = %s",
2162                 evdw_names[fr->vdwtype]);
2163     }
2164     fr->enerdiffsix    = eners[0];
2165     fr->enerdifftwelve = eners[1];
2166     /* The 0.5 is due to the Gromacs definition of the virial */
2167     fr->virdiffsix     = 0.5*virs[0];
2168     fr->virdifftwelve  = 0.5*virs[1];
2169   }
2170 }
2171
2172 void calc_dispcorr(FILE *fplog,t_inputrec *ir,t_forcerec *fr,
2173                    gmx_large_int_t step,int natoms,
2174                    matrix box,real lambda,tensor pres,tensor virial,
2175                    real *prescorr, real *enercorr, real *dvdlcorr)
2176 {
2177     gmx_bool bCorrAll,bCorrPres;
2178     real dvdlambda,invvol,dens,ninter,avcsix,avctwelve,enerdiff,svir=0,spres=0;
2179     int  m;
2180
2181     *prescorr = 0;
2182     *enercorr = 0;
2183     *dvdlcorr = 0;
2184
2185     clear_mat(virial);
2186     clear_mat(pres);
2187
2188     if (ir->eDispCorr != edispcNO) {
2189         bCorrAll  = (ir->eDispCorr == edispcAllEner ||
2190                      ir->eDispCorr == edispcAllEnerPres);
2191         bCorrPres = (ir->eDispCorr == edispcEnerPres ||
2192                      ir->eDispCorr == edispcAllEnerPres);
2193
2194         invvol = 1/det(box);
2195         if (fr->n_tpi)
2196         {
2197             /* Only correct for the interactions with the inserted molecule */
2198             dens = (natoms - fr->n_tpi)*invvol;
2199             ninter = fr->n_tpi;
2200         }
2201         else
2202         {
2203             dens = natoms*invvol;
2204             ninter = 0.5*natoms;
2205         }
2206
2207         if (ir->efep == efepNO)
2208         {
2209             avcsix    = fr->avcsix[0];
2210             avctwelve = fr->avctwelve[0];
2211         }
2212         else
2213         {
2214             avcsix    = (1 - lambda)*fr->avcsix[0]    + lambda*fr->avcsix[1];
2215             avctwelve = (1 - lambda)*fr->avctwelve[0] + lambda*fr->avctwelve[1];
2216         }
2217
2218         enerdiff = ninter*(dens*fr->enerdiffsix - fr->enershiftsix);
2219         *enercorr += avcsix*enerdiff;
2220         dvdlambda = 0.0;
2221         if (ir->efep != efepNO)
2222         {
2223             dvdlambda += (fr->avcsix[1] - fr->avcsix[0])*enerdiff;
2224         }
2225         if (bCorrAll)
2226         {
2227             enerdiff = ninter*(dens*fr->enerdifftwelve - fr->enershifttwelve);
2228             *enercorr += avctwelve*enerdiff;
2229             if (fr->efep != efepNO)
2230             {
2231                 dvdlambda += (fr->avctwelve[1] - fr->avctwelve[0])*enerdiff;
2232             }
2233         }
2234
2235         if (bCorrPres)
2236         {
2237             svir = ninter*dens*avcsix*fr->virdiffsix/3.0;
2238             if (ir->eDispCorr == edispcAllEnerPres)
2239             {
2240                 svir += ninter*dens*avctwelve*fr->virdifftwelve/3.0;
2241             }
2242             /* The factor 2 is because of the Gromacs virial definition */
2243             spres = -2.0*invvol*svir*PRESFAC;
2244
2245             for(m=0; m<DIM; m++) {
2246                 virial[m][m] += svir;
2247                 pres[m][m] += spres;
2248             }
2249             *prescorr += spres;
2250         }
2251
2252         /* Can't currently control when it prints, for now, just print when degugging */
2253         if (debug)
2254         {
2255             if (bCorrAll) {
2256                 fprintf(debug,"Long Range LJ corr.: <C6> %10.4e, <C12> %10.4e\n",
2257                         avcsix,avctwelve);
2258             }
2259             if (bCorrPres)
2260             {
2261                 fprintf(debug,
2262                         "Long Range LJ corr.: Epot %10g, Pres: %10g, Vir: %10g\n",
2263                         *enercorr,spres,svir);
2264             }
2265             else
2266             {
2267                 fprintf(debug,"Long Range LJ corr.: Epot %10g\n",*enercorr);
2268             }
2269         }
2270
2271         if (fr->bSepDVDL && do_per_step(step,ir->nstlog))
2272         {
2273             fprintf(fplog,sepdvdlformat,"Dispersion correction",
2274                     *enercorr,dvdlambda);
2275         }
2276         if (fr->efep != efepNO)
2277         {
2278             *dvdlcorr += dvdlambda;
2279         }
2280     }
2281 }
2282
2283 void do_pbc_first(FILE *fplog,matrix box,t_forcerec *fr,
2284                   t_graph *graph,rvec x[])
2285 {
2286   if (fplog)
2287     fprintf(fplog,"Removing pbc first time\n");
2288   calc_shifts(box,fr->shift_vec);
2289   if (graph) {
2290     mk_mshift(fplog,graph,fr->ePBC,box,x);
2291     if (gmx_debug_at)
2292       p_graph(debug,"do_pbc_first 1",graph);
2293     shift_self(graph,box,x);
2294     /* By doing an extra mk_mshift the molecules that are broken
2295      * because they were e.g. imported from another software
2296      * will be made whole again. Such are the healing powers
2297      * of GROMACS.
2298      */
2299     mk_mshift(fplog,graph,fr->ePBC,box,x);
2300     if (gmx_debug_at)
2301       p_graph(debug,"do_pbc_first 2",graph);
2302   }
2303   if (fplog)
2304     fprintf(fplog,"Done rmpbc\n");
2305 }
2306
2307 static void low_do_pbc_mtop(FILE *fplog,int ePBC,matrix box,
2308                             gmx_mtop_t *mtop,rvec x[],
2309                             gmx_bool bFirst)
2310 {
2311   t_graph *graph;
2312   int mb,as,mol;
2313   gmx_molblock_t *molb;
2314
2315   if (bFirst && fplog)
2316     fprintf(fplog,"Removing pbc first time\n");
2317
2318   snew(graph,1);
2319   as = 0;
2320   for(mb=0; mb<mtop->nmolblock; mb++) {
2321     molb = &mtop->molblock[mb];
2322     if (molb->natoms_mol == 1 ||
2323         (!bFirst && mtop->moltype[molb->type].cgs.nr == 1)) {
2324       /* Just one atom or charge group in the molecule, no PBC required */
2325       as += molb->nmol*molb->natoms_mol;
2326     } else {
2327       /* Pass NULL iso fplog to avoid graph prints for each molecule type */
2328       mk_graph_ilist(NULL,mtop->moltype[molb->type].ilist,
2329                      0,molb->natoms_mol,FALSE,FALSE,graph);
2330
2331       for(mol=0; mol<molb->nmol; mol++) {
2332         mk_mshift(fplog,graph,ePBC,box,x+as);
2333
2334         shift_self(graph,box,x+as);
2335         /* The molecule is whole now.
2336          * We don't need the second mk_mshift call as in do_pbc_first,
2337          * since we no longer need this graph.
2338          */
2339
2340         as += molb->natoms_mol;
2341       }
2342       done_graph(graph);
2343     }
2344   }
2345   sfree(graph);
2346 }
2347
2348 void do_pbc_first_mtop(FILE *fplog,int ePBC,matrix box,
2349                        gmx_mtop_t *mtop,rvec x[])
2350 {
2351   low_do_pbc_mtop(fplog,ePBC,box,mtop,x,TRUE);
2352 }
2353
2354 void do_pbc_mtop(FILE *fplog,int ePBC,matrix box,
2355                  gmx_mtop_t *mtop,rvec x[])
2356 {
2357   low_do_pbc_mtop(fplog,ePBC,box,mtop,x,FALSE);
2358 }
2359
2360 void finish_run(FILE *fplog,t_commrec *cr,const char *confout,
2361                 t_inputrec *inputrec,
2362                 t_nrnb nrnb[],gmx_wallcycle_t wcycle,
2363                 gmx_runtime_t *runtime,
2364                 wallclock_gpu_t *gputimes,
2365                 int omp_nth_pp,
2366                 gmx_bool bWriteStat)
2367 {
2368     int    i,j;
2369     t_nrnb *nrnb_tot=NULL;
2370     real   delta_t;
2371     double nbfs,mflop;
2372
2373     wallcycle_sum(cr,wcycle);
2374
2375     if (cr->nnodes > 1)
2376     {
2377         snew(nrnb_tot,1);
2378 #ifdef GMX_MPI
2379         MPI_Allreduce(nrnb->n,nrnb_tot->n,eNRNB,MPI_DOUBLE,MPI_SUM,
2380                       cr->mpi_comm_mysim);
2381 #endif
2382     }
2383     else
2384     {
2385         nrnb_tot = nrnb;
2386     }
2387
2388 #if defined(GMX_MPI) && !defined(GMX_THREAD_MPI)
2389     if (cr->nnodes > 1)
2390     {
2391         /* reduce nodetime over all MPI processes in the current simulation */
2392         double sum;
2393         MPI_Allreduce(&runtime->proctime,&sum,1,MPI_DOUBLE,MPI_SUM,
2394                       cr->mpi_comm_mysim);
2395         runtime->proctime = sum;
2396     }
2397 #endif
2398
2399     if (SIMMASTER(cr))
2400     {
2401         print_flop(fplog,nrnb_tot,&nbfs,&mflop);
2402     }
2403     if (cr->nnodes > 1)
2404     {
2405         sfree(nrnb_tot);
2406     }
2407
2408     if ((cr->duty & DUTY_PP) && DOMAINDECOMP(cr))
2409     {
2410         print_dd_statistics(cr,inputrec,fplog);
2411     }
2412
2413 #ifdef GMX_MPI
2414     if (PARTDECOMP(cr))
2415     {
2416         if (MASTER(cr))
2417         {
2418             t_nrnb     *nrnb_all;
2419             int        s;
2420             MPI_Status stat;
2421
2422             snew(nrnb_all,cr->nnodes);
2423             nrnb_all[0] = *nrnb;
2424             for(s=1; s<cr->nnodes; s++)
2425             {
2426                 MPI_Recv(nrnb_all[s].n,eNRNB,MPI_DOUBLE,s,0,
2427                          cr->mpi_comm_mysim,&stat);
2428             }
2429             pr_load(fplog,cr,nrnb_all);
2430             sfree(nrnb_all);
2431         }
2432         else
2433         {
2434             MPI_Send(nrnb->n,eNRNB,MPI_DOUBLE,MASTERRANK(cr),0,
2435                      cr->mpi_comm_mysim);
2436         }
2437     }
2438 #endif
2439
2440     if (SIMMASTER(cr))
2441     {
2442         wallcycle_print(fplog,cr->nnodes,cr->npmenodes,runtime->realtime,
2443                         wcycle,gputimes);
2444
2445         if (EI_DYNAMICS(inputrec->eI))
2446         {
2447             delta_t = inputrec->delta_t;
2448         }
2449         else
2450         {
2451             delta_t = 0;
2452         }
2453
2454         if (fplog)
2455         {
2456             print_perf(fplog,runtime->proctime,runtime->realtime,
2457                        cr->nnodes-cr->npmenodes,
2458                        runtime->nsteps_done,delta_t,nbfs,mflop,
2459                        omp_nth_pp);
2460         }
2461         if (bWriteStat)
2462         {
2463             print_perf(stderr,runtime->proctime,runtime->realtime,
2464                        cr->nnodes-cr->npmenodes,
2465                        runtime->nsteps_done,delta_t,nbfs,mflop,
2466                        omp_nth_pp);
2467         }
2468     }
2469 }
2470
2471 extern void initialize_lambdas(FILE *fplog,t_inputrec *ir,int *fep_state,real *lambda,double *lam0)
2472 {
2473     /* this function works, but could probably use a logic rewrite to keep all the different
2474        types of efep straight. */
2475
2476     int i;
2477     t_lambda *fep = ir->fepvals;
2478
2479     if ((ir->efep==efepNO) && (ir->bSimTemp == FALSE)) {
2480         for (i=0;i<efptNR;i++)  {
2481             lambda[i] = 0.0;
2482             if (lam0)
2483             {
2484                 lam0[i] = 0.0;
2485             }
2486         }
2487         return;
2488     } else {
2489         *fep_state = fep->init_fep_state; /* this might overwrite the checkpoint
2490                                              if checkpoint is set -- a kludge is in for now
2491                                              to prevent this.*/
2492         for (i=0;i<efptNR;i++)
2493         {
2494             /* overwrite lambda state with init_lambda for now for backwards compatibility */
2495             if (fep->init_lambda>=0) /* if it's -1, it was never initializd */
2496             {
2497                 lambda[i] = fep->init_lambda;
2498                 if (lam0) {
2499                     lam0[i] = lambda[i];
2500                 }
2501             }
2502             else
2503             {
2504                 lambda[i] = fep->all_lambda[i][*fep_state];
2505                 if (lam0) {
2506                     lam0[i] = lambda[i];
2507                 }
2508             }
2509         }
2510         if (ir->bSimTemp) {
2511             /* need to rescale control temperatures to match current state */
2512             for (i=0;i<ir->opts.ngtc;i++) {
2513                 if (ir->opts.ref_t[i] > 0) {
2514                     ir->opts.ref_t[i] = ir->simtempvals->temperatures[*fep_state];
2515                 }
2516             }
2517         }
2518     }
2519
2520     /* Send to the log the information on the current lambdas */
2521     if (fplog != NULL)
2522     {
2523         fprintf(fplog,"Initial vector of lambda components:[ ");
2524         for (i=0;i<efptNR;i++)
2525         {
2526             fprintf(fplog,"%10.4f ",lambda[i]);
2527         }
2528         fprintf(fplog,"]\n");
2529     }
2530     return;
2531 }
2532
2533
2534 void init_md(FILE *fplog,
2535              t_commrec *cr,t_inputrec *ir,const output_env_t oenv,
2536              double *t,double *t0,
2537              real *lambda, int *fep_state, double *lam0,
2538              t_nrnb *nrnb,gmx_mtop_t *mtop,
2539              gmx_update_t *upd,
2540              int nfile,const t_filenm fnm[],
2541              gmx_mdoutf_t **outf,t_mdebin **mdebin,
2542              tensor force_vir,tensor shake_vir,rvec mu_tot,
2543              gmx_bool *bSimAnn,t_vcm **vcm, t_state *state, unsigned long Flags)
2544 {
2545     int  i,j,n;
2546     real tmpt,mod;
2547
2548     /* Initial values */
2549     *t = *t0       = ir->init_t;
2550
2551     *bSimAnn=FALSE;
2552     for(i=0;i<ir->opts.ngtc;i++)
2553     {
2554         /* set bSimAnn if any group is being annealed */
2555         if(ir->opts.annealing[i]!=eannNO)
2556         {
2557             *bSimAnn = TRUE;
2558         }
2559     }
2560     if (*bSimAnn)
2561     {
2562         update_annealing_target_temp(&(ir->opts),ir->init_t);
2563     }
2564
2565     /* Initialize lambda variables */
2566     initialize_lambdas(fplog,ir,fep_state,lambda,lam0);
2567
2568     if (upd)
2569     {
2570         *upd = init_update(fplog,ir);
2571     }
2572
2573
2574     if (vcm != NULL)
2575     {
2576         *vcm = init_vcm(fplog,&mtop->groups,ir);
2577     }
2578
2579     if (EI_DYNAMICS(ir->eI) && !(Flags & MD_APPENDFILES))
2580     {
2581         if (ir->etc == etcBERENDSEN)
2582         {
2583             please_cite(fplog,"Berendsen84a");
2584         }
2585         if (ir->etc == etcVRESCALE)
2586         {
2587             please_cite(fplog,"Bussi2007a");
2588         }
2589     }
2590
2591     init_nrnb(nrnb);
2592
2593     if (nfile != -1)
2594     {
2595         *outf = init_mdoutf(nfile,fnm,Flags,cr,ir,oenv);
2596
2597         *mdebin = init_mdebin((Flags & MD_APPENDFILES) ? NULL : (*outf)->fp_ene,
2598                               mtop,ir, (*outf)->fp_dhdl);
2599     }
2600
2601     if (ir->bAdress)
2602     {
2603       please_cite(fplog,"Fritsch12");
2604       please_cite(fplog,"Junghans10");
2605     }
2606     /* Initiate variables */
2607     clear_mat(force_vir);
2608     clear_mat(shake_vir);
2609     clear_rvec(mu_tot);
2610
2611     debug_gmx();
2612 }
2613