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, 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 "gromacs/commandline/filenm.h"
  48 #include "gromacs/fileio/gmxfio.h"
  49 #include "gromacs/gmxlib/network.h"
  50 #include "gromacs/mdtypes/commrec.h"
  51 #include "gromacs/utility/binaryinformation.h"
  52 #include "gromacs/utility/cstringutil.h"
  53 #include "gromacs/utility/exceptions.h"
  54 #include "gromacs/utility/fatalerror.h"
  55 #include "gromacs/utility/futil.h"
  56 #include "gromacs/utility/gmxmpi.h"
  57 #include "gromacs/utility/programcontext.h"
  58 #include "gromacs/utility/smalloc.h"
  59 #include "gromacs/utility/snprintf.h"
  60 #include "gromacs/utility/sysinfo.h"
  61
  62 /* The source code in this file should be thread-safe.
  63          Please keep it that way. */
  64
  65 #define BUFSIZE 1024
  66
  67 static void par_fn(char *base, int ftp, const t_commrec *cr,
  68                    gmx_bool bAppendSimId, gmx_bool bAppendNodeId,
  69                    char buf[], int bufsize)
  70 {
  71     if (static_cast<std::size_t>(bufsize) < (std::strlen(base)+10))
  72     {
  73         gmx_mem("Character buffer too small!");
  74     }
  75
  76     /* Copy to buf, and strip extension */
  77     std::strcpy(buf, base);
  78     buf[strlen(base) - std::strlen(ftp2ext(fn2ftp(base))) - 1] = '\0';
  79
  80     if (bAppendSimId)
  81     {
  82         sprintf(buf+strlen(buf), "%d", cr->ms->sim);
  83     }
  84     if (bAppendNodeId)
  85     {
  86         std::strcat(buf, "_rank");
  87         sprintf(buf+strlen(buf), "%d", cr->nodeid);
  88     }
  89     std::strcat(buf, ".");
  90
  91     /* Add extension again */
  92     std::strcat(buf, (ftp == efTPR) ? "tpr" : (ftp == efEDR) ? "edr" : ftp2ext(ftp));
  93     if (debug)
  94     {
  95         fprintf(debug, "rank %d par_fn '%s'\n", cr->nodeid, buf);
  96         if (fn2ftp(buf) == efLOG)
  97         {
  98             fprintf(debug, "log\n");
  99         }
 100     }
 101 }
 102
 103 void check_multi_int(FILE *log, const gmx_multisim_t *ms, int val,
 104                      const char *name,
 105                      gmx_bool bQuiet)
 106 {
 107     int     *ibuf, p;
 108     gmx_bool bCompatible;
 109
 110     if (NULL != log && !bQuiet)
 111     {
 112         fprintf(log, "Multi-checking %s ... ", name);
 113     }
 114
 115     if (ms == NULL)
 116     {
 117         gmx_fatal(FARGS,
 118                   "check_multi_int called with a NULL communication pointer");
 119     }
 120
 121     snew(ibuf, ms->nsim);
 122     ibuf[ms->sim] = val;
 123     gmx_sumi_sim(ms->nsim, ibuf, ms);
 124
 125     bCompatible = TRUE;
 126     for (p = 1; p < ms->nsim; p++)
 127     {
 128         bCompatible = bCompatible && (ibuf[p-1] == ibuf[p]);
 129     }
 130
 131     if (bCompatible)
 132     {
 133         if (NULL != log && !bQuiet)
 134         {
 135             fprintf(log, "OK\n");
 136         }
 137     }
 138     else
 139     {
 140         if (NULL != log)
 141         {
 142             fprintf(log, "\n%s is not equal for all subsystems\n", name);
 143             for (p = 0; p < ms->nsim; p++)
 144             {
 145                 fprintf(log, "  subsystem %d: %d\n", p, ibuf[p]);
 146             }
 147         }
 148         gmx_fatal(FARGS, "The %d subsystems are not compatible\n", ms->nsim);
 149     }
 150
 151     sfree(ibuf);
 152 }
 153
 154 void check_multi_int64(FILE *log, const gmx_multisim_t *ms,
 155                        gmx_int64_t val, const char *name,
 156                        gmx_bool bQuiet)
 157 {
 158     gmx_int64_t      *ibuf;
 159     int               p;
 160     gmx_bool          bCompatible;
 161
 162     if (NULL != log && !bQuiet)
 163     {
 164         fprintf(log, "Multi-checking %s ... ", name);
 165     }
 166
 167     if (ms == NULL)
 168     {
 169         gmx_fatal(FARGS,
 170                   "check_multi_int called with a NULL communication pointer");
 171     }
 172
 173     snew(ibuf, ms->nsim);
 174     ibuf[ms->sim] = val;
 175     gmx_sumli_sim(ms->nsim, ibuf, ms);
 176
 177     bCompatible = TRUE;
 178     for (p = 1; p < ms->nsim; p++)
 179     {
 180         bCompatible = bCompatible && (ibuf[p-1] == ibuf[p]);
 181     }
 182
 183     if (bCompatible)
 184     {
 185         if (NULL != log && !bQuiet)
 186         {
 187             fprintf(log, "OK\n");
 188         }
 189     }
 190     else
 191     {
 192         if (NULL != log)
 193         {
 194             fprintf(log, "\n%s is not equal for all subsystems\n", name);
 195             for (p = 0; p < ms->nsim; p++)
 196             {
 197                 char strbuf[255];
 198                 /* first make the format string */
 199                 snprintf(strbuf, 255, "  subsystem %%d: %s\n",
 200                          "%" GMX_PRId64);
 201                 fprintf(log, strbuf, p, ibuf[p]);
 202             }
 203         }
 204         gmx_fatal(FARGS, "The %d subsystems are not compatible\n", ms->nsim);
 205     }
 206
 207     sfree(ibuf);
 208 }
 209
 210
 211 void gmx_log_open(const char *lognm, const t_commrec *cr,
 212                   gmx_bool bAppendFiles, FILE** fplog)
 213 {
 214     int    pid;
 215     char   host[256];
 216     char   timebuf[STRLEN];
 217     FILE  *fp = *fplog;
 218
 219     if (!bAppendFiles)
 220     {
 221         fp = gmx_fio_fopen(lognm, bAppendFiles ? "a+" : "w+" );
 222     }
 223
 224     gmx_fatal_set_log_file(fp);
 225
 226     /* Get some machine parameters */
 227     gmx_gethostname(host, 256);
 228     pid = gmx_getpid();
 229     gmx_format_current_time(timebuf, STRLEN);
 230
 231     if (bAppendFiles)
 232     {
 233         fprintf(fp,
 234                 "\n"
 235                 "\n"
 236                 "-----------------------------------------------------------\n"
 237                 "Restarting from checkpoint, appending to previous log file.\n"
 238                 "\n"
 239                 );
 240     }
 241
 242     fprintf(fp,
 243             "Log file opened on %s"
 244             "Host: %s  pid: %d  rank ID: %d  number of ranks:  %d\n",
 245             timebuf, host, pid, cr->nodeid, cr->nnodes);
 246     try
 247     {
 248         gmx::BinaryInformationSettings settings;
 249         settings.extendedInfo(true);
 250         settings.copyright(!bAppendFiles);
 251         gmx::printBinaryInformation(fp, gmx::getProgramContext(), settings);
 252     }
 253     GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR;
 254     fprintf(fp, "\n");
 255
 256     fflush(fp);
 257
 258     *fplog = fp;
 259 }
 260
 261 void gmx_log_close(FILE *fp)
 262 {
 263     if (fp)
 264     {
 265         gmx_fatal_set_log_file(NULL);
 266         gmx_fio_fclose(fp);
 267     }
 268 }
 269
 270 void init_multisystem(t_commrec *cr, int nsim, char **multidirs,
 271                       int nfile, const t_filenm fnm[], gmx_bool bParFn)
 272 {
 273     gmx_multisim_t *ms;
 274     int             nnodes, nnodpersim, sim, i, ftp;
 275     char            buf[256];
 276 #if GMX_MPI
 277     MPI_Group       mpi_group_world;
 278     int            *rank;
 279 #endif
 280
 281 #if !GMX_MPI
 282     if (nsim > 1)
 283     {
 284         gmx_fatal(FARGS, "This binary is compiled without MPI support, can not do multiple simulations.");
 285     }
 286 #endif
 287
 288     nnodes  = cr->nnodes;
 289     if (nnodes % nsim != 0)
 290     {
 291         gmx_fatal(FARGS, "The number of ranks (%d) is not a multiple of the number of simulations (%d)", nnodes, nsim);
 292     }
 293
 294     nnodpersim = nnodes/nsim;
 295     sim        = cr->nodeid/nnodpersim;
 296
 297     if (debug)
 298     {
 299         fprintf(debug, "We have %d simulations, %d ranks per simulation, local simulation is %d\n", nsim, nnodpersim, sim);
 300     }
 301
 302     snew(ms, 1);
 303     cr->ms   = ms;
 304     ms->nsim = nsim;
 305     ms->sim  = sim;
 306 #if GMX_MPI
 307     /* Create a communicator for the master nodes */
 308     snew(rank, ms->nsim);
 309     for (i = 0; i < ms->nsim; i++)
 310     {
 311         rank[i] = i*nnodpersim;
 312     }
 313     MPI_Comm_group(MPI_COMM_WORLD, &mpi_group_world);
 314     MPI_Group_incl(mpi_group_world, nsim, rank, &ms->mpi_group_masters);
 315     sfree(rank);
 316     MPI_Comm_create(MPI_COMM_WORLD, ms->mpi_group_masters,
 317                     &ms->mpi_comm_masters);
 318
 319 #if !MPI_IN_PLACE_EXISTS
 320     /* initialize the MPI_IN_PLACE replacement buffers */
 321     snew(ms->mpb, 1);
 322     ms->mpb->ibuf        = NULL;
 323     ms->mpb->libuf       = NULL;
 324     ms->mpb->fbuf        = NULL;
 325     ms->mpb->dbuf        = NULL;
 326     ms->mpb->ibuf_alloc  = 0;
 327     ms->mpb->libuf_alloc = 0;
 328     ms->mpb->fbuf_alloc  = 0;
 329     ms->mpb->dbuf_alloc  = 0;
 330 #endif
 331
 332 #endif
 333
 334     /* Reduce the intra-simulation communication */
 335     cr->sim_nodeid = cr->nodeid % nnodpersim;
 336     cr->nnodes     = nnodpersim;
 337 #if GMX_MPI
 338     MPI_Comm_split(MPI_COMM_WORLD, sim, cr->sim_nodeid, &cr->mpi_comm_mysim);
 339     cr->mpi_comm_mygroup = cr->mpi_comm_mysim;
 340     cr->nodeid           = cr->sim_nodeid;
 341 #endif
 342
 343     if (debug)
 344     {
 345         fprintf(debug, "This is simulation %d", cr->ms->sim);
 346         if (PAR(cr))
 347         {
 348             fprintf(debug, ", local number of ranks %d, local rank ID %d",
 349                     cr->nnodes, cr->sim_nodeid);
 350         }
 351         fprintf(debug, "\n\n");
 352     }
 353
 354     if (multidirs)
 355     {
 356         if (debug)
 357         {
 358             fprintf(debug, "Changing to directory %s\n", multidirs[cr->ms->sim]);
 359         }
 360         gmx_chdir(multidirs[cr->ms->sim]);
 361     }
 362     else if (bParFn)
 363     {
 364         /* Patch output and tpx, cpt and rerun input file names */
 365         for (i = 0; (i < nfile); i++)
 366         {
 367             /* Because of possible multiple extensions per type we must look
 368              * at the actual file name
 369              */
 370             if (is_output(&fnm[i]) ||
 371                 fnm[i].ftp == efTPR || fnm[i].ftp == efCPT ||
 372                 strcmp(fnm[i].opt, "-rerun") == 0)
 373             {
 374                 ftp = fn2ftp(fnm[i].fns[0]);
 375                 par_fn(fnm[i].fns[0], ftp, cr, TRUE, FALSE, buf, 255);
 376                 sfree(fnm[i].fns[0]);
 377                 fnm[i].fns[0] = gmx_strdup(buf);
 378             }
 379         }
 380     }
 381 }