src/gromacs/mdlib/main.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
   5  * Copyright (c) 2001-2004, The GROMACS development team.
   6  * Copyright (c) 2013,2014,2015,2016,2017, by the GROMACS development team, led by
   7  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   8  * and including many others, as listed in the AUTHORS file in the
   9  * top-level source directory and at http://www.gromacs.org.
  10  *
  11  * GROMACS is free software; you can redistribute it and/or
  12  * modify it under the terms of the GNU Lesser General Public License
  13  * as published by the Free Software Foundation; either version 2.1
  14  * of the License, or (at your option) any later version.
  15  *
  16  * GROMACS is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19  * Lesser General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU Lesser General Public
  22  * License along with GROMACS; if not, see
  23  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  24  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  25  *
  26  * If you want to redistribute modifications to GROMACS, please
  27  * consider that scientific software is very special. Version
  28  * control is crucial - bugs must be traceable. We will be happy to
  29  * consider code for inclusion in the official distribution, but
  30  * derived work must not be called official GROMACS. Details are found
  31  * in the README & COPYING files - if they are missing, get the
  32  * official version at http://www.gromacs.org.
  33  *
  34  * To help us fund GROMACS development, we humbly ask that you cite
  35  * the research papers on the package. Check out http://www.gromacs.org.
  36  */
  37 #include "gmxpre.h"
  38
  39 #include "main.h"
  40
  41 #include "config.h"
  42
  43 #include <cstdio>
  44 #include <cstdlib>
  45 #include <cstring>
  46
  47 #include <string>
  48
  49 #include "gromacs/commandline/filenm.h"
  50 #include "gromacs/fileio/gmxfio.h"
  51 #include "gromacs/gmxlib/network.h"
  52 #include "gromacs/mdtypes/commrec.h"
  53 #include "gromacs/utility/binaryinformation.h"
  54 #include "gromacs/utility/cstringutil.h"
  55 #include "gromacs/utility/exceptions.h"
  56 #include "gromacs/utility/fatalerror.h"
  57 #include "gromacs/utility/futil.h"
  58 #include "gromacs/utility/gmxmpi.h"
  59 #include "gromacs/utility/path.h"
  60 #include "gromacs/utility/programcontext.h"
  61 #include "gromacs/utility/smalloc.h"
  62 #include "gromacs/utility/snprintf.h"
  63 #include "gromacs/utility/stringutil.h"
  64 #include "gromacs/utility/sysinfo.h"
  65
  66 /* The source code in this file should be thread-safe.
  67          Please keep it that way. */
  68
  69 // TODO move this to multi-sim module
  70 void check_multi_int(FILE *log, const gmx_multisim_t *ms, int val,
  71                      const char *name,
  72                      gmx_bool bQuiet)
  73 {
  74     int     *ibuf, p;
  75     gmx_bool bCompatible;
  76
  77     if (nullptr != log && !bQuiet)
  78     {
  79         fprintf(log, "Multi-checking %s ... ", name);
  80     }
  81
  82     if (ms == nullptr)
  83     {
  84         gmx_fatal(FARGS,
  85                   "check_multi_int called with a NULL communication pointer");
  86     }
  87
  88     snew(ibuf, ms->nsim);
  89     ibuf[ms->sim] = val;
  90     gmx_sumi_sim(ms->nsim, ibuf, ms);
  91
  92     bCompatible = TRUE;
  93     for (p = 1; p < ms->nsim; p++)
  94     {
  95         bCompatible = bCompatible && (ibuf[p-1] == ibuf[p]);
  96     }
  97
  98     if (bCompatible)
  99     {
 100         if (nullptr != log && !bQuiet)
 101         {
 102             fprintf(log, "OK\n");
 103         }
 104     }
 105     else
 106     {
 107         if (nullptr != log)
 108         {
 109             fprintf(log, "\n%s is not equal for all subsystems\n", name);
 110             for (p = 0; p < ms->nsim; p++)
 111             {
 112                 fprintf(log, "  subsystem %d: %d\n", p, ibuf[p]);
 113             }
 114         }
 115         gmx_fatal(FARGS, "The %d subsystems are not compatible\n", ms->nsim);
 116     }
 117
 118     sfree(ibuf);
 119 }
 120
 121 // TODO move this to multi-sim module
 122 void check_multi_int64(FILE *log, const gmx_multisim_t *ms,
 123                        gmx_int64_t val, const char *name,
 124                        gmx_bool bQuiet)
 125 {
 126     gmx_int64_t      *ibuf;
 127     int               p;
 128     gmx_bool          bCompatible;
 129
 130     if (nullptr != log && !bQuiet)
 131     {
 132         fprintf(log, "Multi-checking %s ... ", name);
 133     }
 134
 135     if (ms == nullptr)
 136     {
 137         gmx_fatal(FARGS,
 138                   "check_multi_int called with a NULL communication pointer");
 139     }
 140
 141     snew(ibuf, ms->nsim);
 142     ibuf[ms->sim] = val;
 143     gmx_sumli_sim(ms->nsim, ibuf, ms);
 144
 145     bCompatible = TRUE;
 146     for (p = 1; p < ms->nsim; p++)
 147     {
 148         bCompatible = bCompatible && (ibuf[p-1] == ibuf[p]);
 149     }
 150
 151     if (bCompatible)
 152     {
 153         if (nullptr != log && !bQuiet)
 154         {
 155             fprintf(log, "OK\n");
 156         }
 157     }
 158     else
 159     {
 160         // TODO Part of this error message would also be good to go to
 161         // stderr (from one rank of one sim only)
 162         if (nullptr != log)
 163         {
 164             fprintf(log, "\n%s is not equal for all subsystems\n", name);
 165             for (p = 0; p < ms->nsim; p++)
 166             {
 167                 char strbuf[255];
 168                 /* first make the format string */
 169                 snprintf(strbuf, 255, "  subsystem %%d: %s\n",
 170                          "%" GMX_PRId64);
 171                 fprintf(log, strbuf, p, ibuf[p]);
 172             }
 173         }
 174         gmx_fatal(FARGS, "The %d subsystems are not compatible\n", ms->nsim);
 175     }
 176
 177     sfree(ibuf);
 178 }
 179
 180
 181 void gmx_log_open(const char *lognm, const t_commrec *cr,
 182                   gmx_bool bAppendFiles, FILE** fplog)
 183 {
 184     int    pid;
 185     char   host[256];
 186     char   timebuf[STRLEN];
 187     FILE  *fp = *fplog;
 188
 189     if (!bAppendFiles)
 190     {
 191         fp = gmx_fio_fopen(lognm, bAppendFiles ? "a+" : "w+" );
 192     }
 193
 194     gmx_fatal_set_log_file(fp);
 195
 196     /* Get some machine parameters */
 197     gmx_gethostname(host, 256);
 198     pid = gmx_getpid();
 199     gmx_format_current_time(timebuf, STRLEN);
 200
 201     if (bAppendFiles)
 202     {
 203         fprintf(fp,
 204                 "\n"
 205                 "\n"
 206                 "-----------------------------------------------------------\n"
 207                 "Restarting from checkpoint, appending to previous log file.\n"
 208                 "\n"
 209                 );
 210     }
 211
 212     fprintf(fp,
 213             "Log file opened on %s"
 214             "Host: %s  pid: %d  rank ID: %d  number of ranks:  %d\n",
 215             timebuf, host, pid, cr->nodeid, cr->nnodes);
 216     try
 217     {
 218         gmx::BinaryInformationSettings settings;
 219         settings.extendedInfo(true);
 220         settings.copyright(!bAppendFiles);
 221         gmx::printBinaryInformation(fp, gmx::getProgramContext(), settings);
 222     }
 223     GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR;
 224     fprintf(fp, "\n");
 225
 226     fflush(fp);
 227
 228     *fplog = fp;
 229 }
 230
 231 void gmx_log_close(FILE *fp)
 232 {
 233     if (fp)
 234     {
 235         gmx_fatal_set_log_file(nullptr);
 236         gmx_fio_fclose(fp);
 237     }
 238 }
 239
 240 // TODO move this to multi-sim module
 241 void init_multisystem(t_commrec *cr, int nsim, char **multidirs,
 242                       int nfile, const t_filenm fnm[])
 243 {
 244     gmx_multisim_t *ms;
 245     int             nnodes, nnodpersim, sim, i;
 246 #if GMX_MPI
 247     MPI_Group       mpi_group_world;
 248     int            *rank;
 249 #endif
 250
 251 #if !GMX_MPI
 252     if (nsim > 1)
 253     {
 254         gmx_fatal(FARGS, "This binary is compiled without MPI support, can not do multiple simulations.");
 255     }
 256 #endif
 257
 258     nnodes  = cr->nnodes;
 259     if (nnodes % nsim != 0)
 260     {
 261         gmx_fatal(FARGS, "The number of ranks (%d) is not a multiple of the number of simulations (%d)", nnodes, nsim);
 262     }
 263
 264     nnodpersim = nnodes/nsim;
 265     sim        = cr->nodeid/nnodpersim;
 266
 267     if (debug)
 268     {
 269         fprintf(debug, "We have %d simulations, %d ranks per simulation, local simulation is %d\n", nsim, nnodpersim, sim);
 270     }
 271
 272     snew(ms, 1);
 273     cr->ms   = ms;
 274     ms->nsim = nsim;
 275     ms->sim  = sim;
 276 #if GMX_MPI
 277     /* Create a communicator for the master nodes */
 278     snew(rank, ms->nsim);
 279     for (i = 0; i < ms->nsim; i++)
 280     {
 281         rank[i] = i*nnodpersim;
 282     }
 283     MPI_Comm_group(MPI_COMM_WORLD, &mpi_group_world);
 284     MPI_Group_incl(mpi_group_world, nsim, rank, &ms->mpi_group_masters);
 285     sfree(rank);
 286     MPI_Comm_create(MPI_COMM_WORLD, ms->mpi_group_masters,
 287                     &ms->mpi_comm_masters);
 288
 289 #if !MPI_IN_PLACE_EXISTS
 290     /* initialize the MPI_IN_PLACE replacement buffers */
 291     snew(ms->mpb, 1);
 292     ms->mpb->ibuf        = NULL;
 293     ms->mpb->libuf       = NULL;
 294     ms->mpb->fbuf        = NULL;
 295     ms->mpb->dbuf        = NULL;
 296     ms->mpb->ibuf_alloc  = 0;
 297     ms->mpb->libuf_alloc = 0;
 298     ms->mpb->fbuf_alloc  = 0;
 299     ms->mpb->dbuf_alloc  = 0;
 300 #endif
 301
 302 #endif
 303
 304     /* Reduce the intra-simulation communication */
 305     cr->sim_nodeid = cr->nodeid % nnodpersim;
 306     cr->nnodes     = nnodpersim;
 307 #if GMX_MPI
 308     MPI_Comm_split(MPI_COMM_WORLD, sim, cr->sim_nodeid, &cr->mpi_comm_mysim);
 309     cr->mpi_comm_mygroup = cr->mpi_comm_mysim;
 310     cr->nodeid           = cr->sim_nodeid;
 311 #endif
 312
 313     if (debug)
 314     {
 315         fprintf(debug, "This is simulation %d", cr->ms->sim);
 316         if (PAR(cr))
 317         {
 318             fprintf(debug, ", local number of ranks %d, local rank ID %d",
 319                     cr->nnodes, cr->sim_nodeid);
 320         }
 321         fprintf(debug, "\n\n");
 322     }
 323
 324     if (multidirs)
 325     {
 326         if (debug)
 327         {
 328             fprintf(debug, "Changing to directory %s\n", multidirs[cr->ms->sim]);
 329         }
 330         gmx_chdir(multidirs[cr->ms->sim]);
 331     }
 332     else
 333     {
 334         try
 335         {
 336             std::string rankString = gmx::formatString("%d", cr->ms->sim);
 337             /* Patch output and tpx, cpt and rerun input file names */
 338             for (i = 0; (i < nfile); i++)
 339             {
 340                 /* Because of possible multiple extensions per type we must look
 341                  * at the actual file name for rerun. */
 342                 if (is_output(&fnm[i]) ||
 343                     fnm[i].ftp == efTPR || fnm[i].ftp == efCPT ||
 344                     strcmp(fnm[i].opt, "-rerun") == 0)
 345                 {
 346                     std::string newFileName = gmx::Path::concatenateBeforeExtension(fnm[i].fns[0], rankString);
 347                     sfree(fnm[i].fns[0]);
 348                     fnm[i].fns[0] = gmx_strdup(newFileName.c_str());
 349                 }
 350             }
 351         }
 352         catch (gmx::GromacsException &e)
 353         {
 354             e.prependContext(gmx::formatString("Failed to modify mdrun -multi filename to add per-simulation suffix. You could perhaps reorganize your files and try mdrun -multidir.\n"));
 355             throw;
 356         }
 357     }
 358 }