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) 2013,2014,2015,2016,2017,2018, 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/gmxassert.h"
  59 #include "gromacs/utility/gmxmpi.h"
  60 #include "gromacs/utility/path.h"
  61 #include "gromacs/utility/programcontext.h"
  62 #include "gromacs/utility/smalloc.h"
  63 #include "gromacs/utility/snprintf.h"
  64 #include "gromacs/utility/stringutil.h"
  65 #include "gromacs/utility/sysinfo.h"
  66
  67 /* The source code in this file should be thread-safe.
  68          Please keep it that way. */
  69
  70 // TODO move this to multi-sim module
  71 void check_multi_int(FILE *log, const gmx_multisim_t *ms, int val,
  72                      const char *name,
  73                      gmx_bool bQuiet)
  74 {
  75     int     *ibuf, p;
  76     gmx_bool bCompatible;
  77
  78     if (nullptr != log && !bQuiet)
  79     {
  80         fprintf(log, "Multi-checking %s ... ", name);
  81     }
  82
  83     if (ms == nullptr)
  84     {
  85         gmx_fatal(FARGS,
  86                   "check_multi_int called with a NULL communication pointer");
  87     }
  88
  89     snew(ibuf, ms->nsim);
  90     ibuf[ms->sim] = val;
  91     gmx_sumi_sim(ms->nsim, ibuf, ms);
  92
  93     bCompatible = TRUE;
  94     for (p = 1; p < ms->nsim; p++)
  95     {
  96         bCompatible = bCompatible && (ibuf[p-1] == ibuf[p]);
  97     }
  98
  99     if (bCompatible)
 100     {
 101         if (nullptr != log && !bQuiet)
 102         {
 103             fprintf(log, "OK\n");
 104         }
 105     }
 106     else
 107     {
 108         if (nullptr != log)
 109         {
 110             fprintf(log, "\n%s is not equal for all subsystems\n", name);
 111             for (p = 0; p < ms->nsim; p++)
 112             {
 113                 fprintf(log, "  subsystem %d: %d\n", p, ibuf[p]);
 114             }
 115         }
 116         gmx_fatal(FARGS, "The %d subsystems are not compatible\n", ms->nsim);
 117     }
 118
 119     sfree(ibuf);
 120 }
 121
 122 // TODO move this to multi-sim module
 123 void check_multi_int64(FILE *log, const gmx_multisim_t *ms,
 124                        int64_t val, const char *name,
 125                        gmx_bool bQuiet)
 126 {
 127     int64_t          *ibuf;
 128     int               p;
 129     gmx_bool          bCompatible;
 130
 131     if (nullptr != log && !bQuiet)
 132     {
 133         fprintf(log, "Multi-checking %s ... ", name);
 134     }
 135
 136     if (ms == nullptr)
 137     {
 138         gmx_fatal(FARGS,
 139                   "check_multi_int called with a NULL communication pointer");
 140     }
 141
 142     snew(ibuf, ms->nsim);
 143     ibuf[ms->sim] = val;
 144     gmx_sumli_sim(ms->nsim, ibuf, ms);
 145
 146     bCompatible = TRUE;
 147     for (p = 1; p < ms->nsim; p++)
 148     {
 149         bCompatible = bCompatible && (ibuf[p-1] == ibuf[p]);
 150     }
 151
 152     if (bCompatible)
 153     {
 154         if (nullptr != log && !bQuiet)
 155         {
 156             fprintf(log, "OK\n");
 157         }
 158     }
 159     else
 160     {
 161         // TODO Part of this error message would also be good to go to
 162         // stderr (from one rank of one sim only)
 163         if (nullptr != log)
 164         {
 165             fprintf(log, "\n%s is not equal for all subsystems\n", name);
 166             for (p = 0; p < ms->nsim; p++)
 167             {
 168                 char strbuf[255];
 169                 /* first make the format string */
 170                 snprintf(strbuf, 255, "  subsystem %%d: %s\n",
 171                          "%" PRId64);
 172                 fprintf(log, strbuf, p, ibuf[p]);
 173             }
 174         }
 175         gmx_fatal(FARGS, "The %d subsystems are not compatible\n", ms->nsim);
 176     }
 177
 178     sfree(ibuf);
 179 }
 180
 181
 182 void gmx_log_open(const char *lognm, const t_commrec *cr,
 183                   gmx_bool bAppendFiles, FILE** fplog)
 184 {
 185     int    pid;
 186     char   host[256];
 187     char   timebuf[STRLEN];
 188     FILE  *fp = *fplog;
 189
 190     if (!bAppendFiles)
 191     {
 192         fp = gmx_fio_fopen(lognm, bAppendFiles ? "a+" : "w+" );
 193     }
 194
 195     gmx_fatal_set_log_file(fp);
 196
 197     /* Get some machine parameters */
 198     gmx_gethostname(host, 256);
 199     pid = gmx_getpid();
 200     gmx_format_current_time(timebuf, STRLEN);
 201
 202     if (bAppendFiles)
 203     {
 204         fprintf(fp,
 205                 "\n"
 206                 "\n"
 207                 "-----------------------------------------------------------\n"
 208                 "Restarting from checkpoint, appending to previous log file.\n"
 209                 "\n"
 210                 );
 211     }
 212
 213     fprintf(fp,
 214             "Log file opened on %s"
 215             "Host: %s  pid: %d  rank ID: %d  number of ranks:  %d\n",
 216             timebuf, host, pid, cr->nodeid, cr->nnodes);
 217     try
 218     {
 219         gmx::BinaryInformationSettings settings;
 220         settings.extendedInfo(true);
 221         settings.copyright(!bAppendFiles);
 222         gmx::printBinaryInformation(fp, gmx::getProgramContext(), settings);
 223     }
 224     GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR;
 225     fprintf(fp, "\n");
 226
 227     fflush(fp);
 228
 229     *fplog = fp;
 230 }
 231
 232 void gmx_log_close(FILE *fp)
 233 {
 234     if (fp)
 235     {
 236         gmx_fatal_set_log_file(nullptr);
 237         gmx_fio_fclose(fp);
 238     }
 239 }
 240
 241 // TODO move this to multi-sim module
 242 gmx_multisim_t *init_multisystem(MPI_Comm                         comm,
 243                                  gmx::ArrayRef<const std::string> multidirs)
 244 {
 245     gmx_multisim_t *ms;
 246 #if GMX_MPI
 247     MPI_Group       mpi_group_world;
 248     int            *rank;
 249 #endif
 250
 251     if (multidirs.empty())
 252     {
 253         return nullptr;
 254     }
 255
 256     if (!GMX_LIB_MPI && !multidirs.empty())
 257     {
 258         gmx_fatal(FARGS, "mdrun -multidir is only supported when GROMACS has been "
 259                   "configured with a proper external MPI library.");
 260     }
 261
 262     if (multidirs.size() == 1)
 263     {
 264         /* NOTE: It would be nice if this special case worked, but this requires checks/tests. */
 265         gmx_fatal(FARGS, "To run mdrun in multiple simulation mode, more then one "
 266                   "actual simulation is required. The single simulation case is not supported.");
 267     }
 268
 269 #if GMX_MPI
 270     int numRanks;
 271     MPI_Comm_size(comm, &numRanks);
 272     if (numRanks % multidirs.size() != 0)
 273     {
 274         gmx_fatal(FARGS, "The number of ranks (%d) is not a multiple of the number of simulations (%td)", numRanks, multidirs.size());
 275     }
 276
 277     int numRanksPerSim = numRanks/multidirs.size();
 278     int rankWithinComm;
 279     MPI_Comm_rank(comm, &rankWithinComm);
 280
 281     if (debug)
 282     {
 283         fprintf(debug, "We have %td simulations, %d ranks per simulation, local simulation is %d\n", multidirs.size(), numRanksPerSim, rankWithinComm/numRanksPerSim);
 284     }
 285
 286     ms       = new gmx_multisim_t;
 287     ms->nsim = multidirs.size();
 288     ms->sim  = rankWithinComm/numRanksPerSim;
 289     /* Create a communicator for the master nodes */
 290     snew(rank, ms->nsim);
 291     for (int i = 0; i < ms->nsim; i++)
 292     {
 293         rank[i] = i*numRanksPerSim;
 294     }
 295     MPI_Comm_group(comm, &mpi_group_world);
 296     MPI_Group_incl(mpi_group_world, ms->nsim, rank, &ms->mpi_group_masters);
 297     sfree(rank);
 298     MPI_Comm_create(MPI_COMM_WORLD, ms->mpi_group_masters,
 299                     &ms->mpi_comm_masters);
 300
 301 #if !MPI_IN_PLACE_EXISTS
 302     /* initialize the MPI_IN_PLACE replacement buffers */
 303     snew(ms->mpb, 1);
 304     ms->mpb->ibuf        = NULL;
 305     ms->mpb->libuf       = NULL;
 306     ms->mpb->fbuf        = NULL;
 307     ms->mpb->dbuf        = NULL;
 308     ms->mpb->ibuf_alloc  = 0;
 309     ms->mpb->libuf_alloc = 0;
 310     ms->mpb->fbuf_alloc  = 0;
 311     ms->mpb->dbuf_alloc  = 0;
 312 #endif
 313
 314     // TODO This should throw upon error
 315     gmx_chdir(multidirs[ms->sim].c_str());
 316 #else
 317     GMX_UNUSED_VALUE(comm);
 318     ms = nullptr;
 319 #endif
 320
 321     return ms;
 322 }
 323
 324 void done_multisim(gmx_multisim_t *ms)
 325 {
 326     if (nullptr != ms)
 327     {
 328         done_mpi_in_place_buf(ms->mpb);
 329         delete ms;
 330     }
 331 }