src/gromacs/mdlib/stophandler.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2018,2019, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*! \internal \file
  36  * \brief Defines StopHandler, a helper class and two stop conditions.
  37  *
  38  * \author Pascal Merz <pascal.merz@colorado.edu>
  39  * \ingroup module_mdlib
  40  */
  41 #include "gmxpre.h"
  42
  43 #include "stophandler.h"
  44
  45 #include "config.h"
  46
  47 #include <memory>
  48
  49 #include "gromacs/timing/walltime_accounting.h"
  50 #include "gromacs/utility/cstringutil.h"
  51
  52 namespace gmx
  53 {
  54
  55 StopHandler::StopHandler(compat::not_null<SimulationSignal*>      signal,
  56                          bool                                     simulationShareState,
  57                          std::vector<std::function<StopSignal()>> stopConditions,
  58                          bool                                     neverUpdateNeighborList) :
  59     signal_(*signal),
  60     stopConditions_(std::move(stopConditions)),
  61     neverUpdateNeighborlist_(neverUpdateNeighborList)
  62 {
  63     if (simulationShareState)
  64     {
  65         signal_.isLocal = false;
  66     }
  67 }
  68
  69 StopConditionSignal::StopConditionSignal(int nstList, bool makeBinaryReproducibleSimulation, int nstSignalComm) :
  70     handledStopCondition_(gmx_stop_cond_none),
  71     makeBinaryReproducibleSimulation_(makeBinaryReproducibleSimulation),
  72     nstSignalComm_(nstSignalComm),
  73     nstList_(nstList)
  74 {
  75 }
  76
  77 StopSignal StopConditionSignal::getSignal(FILE* fplog)
  78 {
  79     StopSignal signal = StopSignal::noSignal;
  80
  81     /* Check whether everything is still alright */
  82     if (static_cast<int>(gmx_get_stop_condition()) > handledStopCondition_)
  83     {
  84         int nsteps_stop = -1;
  85
  86         /* this just makes signals[].sig compatible with the hack
  87            of sending signals around by MPI_Reduce together with
  88            other floats */
  89         if ((gmx_get_stop_condition() == gmx_stop_cond_next_ns)
  90             || (makeBinaryReproducibleSimulation_ && gmx_get_stop_condition() == gmx_stop_cond_next))
  91         {
  92             /* We need at least two global communication steps to pass
  93              * around the signal. We stop at a pair-list creation step
  94              * to allow for exact continuation, when possible.
  95              */
  96             signal      = StopSignal::stopAtNextNSStep;
  97             nsteps_stop = std::max(nstList_, 2 * nstSignalComm_);
  98         }
  99         else if (gmx_get_stop_condition() == gmx_stop_cond_next)
 100         {
 101             /* Stop directly after the next global communication step.
 102              * This breaks exact continuation.
 103              */
 104             signal      = StopSignal::stopImmediately;
 105             nsteps_stop = nstSignalComm_ + 1;
 106         }
 107         if (fplog)
 108         {
 109             fprintf(fplog, "\n\nReceived the %s signal, stopping within %d steps\n\n",
 110                     gmx_get_signal_name(), nsteps_stop);
 111             fflush(fplog);
 112         }
 113         fprintf(stderr, "\n\nReceived the %s signal, stopping within %d steps\n\n",
 114                 gmx_get_signal_name(), nsteps_stop);
 115         fflush(stderr);
 116         handledStopCondition_ = static_cast<int>(gmx_get_stop_condition());
 117     }
 118
 119     return signal;
 120 }
 121
 122 StopConditionTime::StopConditionTime(int nstList, real maximumHoursToRun, int nstSignalComm) :
 123     signalSent_(false),
 124     maximumHoursToRun_(maximumHoursToRun),
 125     nstList_(nstList),
 126     nstSignalComm_(nstSignalComm),
 127     neverUpdateNeighborlist_(nstList <= 0)
 128 {
 129 }
 130
 131 StopSignal StopConditionTime::getSignal(bool bNS, int64_t step, FILE* fplog, gmx_walltime_accounting_t walltime_accounting)
 132 {
 133     if (signalSent_)
 134     {
 135         // We only want to send it once, but might be called again before run is terminated
 136         return StopSignal::noSignal;
 137     }
 138     if ((bNS || neverUpdateNeighborlist_)
 139         && walltime_accounting_get_time_since_start(walltime_accounting)
 140                    > maximumHoursToRun_ * 60.0 * 60.0 * 0.99)
 141     {
 142         /* Signal to terminate the run */
 143         char sbuf[STEPSTRSIZE];
 144         int  nsteps_stop = std::max(nstList_, 2 * nstSignalComm_);
 145         if (fplog)
 146         {
 147             fprintf(fplog,
 148                     "\nStep %s: Run time exceeded %.3f hours, "
 149                     "will terminate the run within %d steps\n",
 150                     gmx_step_str(step, sbuf), maximumHoursToRun_ * 0.99, nsteps_stop);
 151         }
 152         fprintf(stderr,
 153                 "\nStep %s: Run time exceeded %.3f hours, "
 154                 "will terminate the run within %d steps\n",
 155                 gmx_step_str(step, sbuf), maximumHoursToRun_ * 0.99, nsteps_stop);
 156         signalSent_ = true;
 157         return StopSignal::stopAtNextNSStep;
 158     }
 159     return StopSignal::noSignal;
 160 }
 161
 162 void StopHandlerBuilder::registerStopCondition(std::function<StopSignal()> stopCondition)
 163 {
 164     stopConditions_.emplace_back(std::move(stopCondition));
 165 };
 166
 167 std::unique_ptr<StopHandler> StopHandlerBuilder::getStopHandlerMD(compat::not_null<SimulationSignal*> signal,
 168                                                                   bool simulationShareState,
 169                                                                   bool isMaster,
 170                                                                   int  nstList,
 171                                                                   bool makeBinaryReproducibleSimulation,
 172                                                                   int   nstSignalComm,
 173                                                                   real  maximumHoursToRun,
 174                                                                   bool  neverUpdateNeighborList,
 175                                                                   FILE* fplog,
 176                                                                   const int64_t&  step,
 177                                                                   const gmx_bool& bNS,
 178                                                                   gmx_walltime_accounting_t walltime_accounting)
 179 {
 180     if (!GMX_THREAD_MPI || isMaster)
 181     {
 182         // Using shared ptr because move-only callable not supported by std::function.
 183         // Would require replacement such as fu2::function or cxx_function.
 184         auto stopConditionSignal = std::make_shared<StopConditionSignal>(
 185                 nstList, makeBinaryReproducibleSimulation, nstSignalComm);
 186         registerStopCondition(
 187                 [stopConditionSignal, fplog]() { return stopConditionSignal->getSignal(fplog); });
 188     }
 189
 190     if (isMaster && maximumHoursToRun > 0)
 191     {
 192         auto stopConditionTime =
 193                 std::make_shared<StopConditionTime>(nstList, maximumHoursToRun, nstSignalComm);
 194         registerStopCondition([stopConditionTime, &bNS, &step, fplog, walltime_accounting]() {
 195             return stopConditionTime->getSignal(bNS, step, fplog, walltime_accounting);
 196         });
 197     }
 198
 199     return std::make_unique<StopHandler>(signal, simulationShareState, stopConditions_,
 200                                          neverUpdateNeighborList);
 201 }
 202
 203 } // namespace gmx