src/gromacs/mdrun/simulationcontext.h

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2018,2019,2020, 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 #ifndef GROMACS_SIMULATIONCONTEXT_H
  36 #define GROMACS_SIMULATIONCONTEXT_H
  37
  38 /*! \libinternal
  39  * \file
  40  * \brief Provide ways for client code to own simulation resources.
  41  *
  42  * For `gmx mdrun` to be implemented as a client program, public API needs to
  43  * provide a way to create and manipulate the SimulationContext.
  44  *
  45  * \author M. Eric Irrgang <ericirrgang@gmail.com>
  46  * \ingroup module_mdrun
  47  * \inlibraryapi
  48  */
  49
  50 #include <memory>
  51 #include <string>
  52
  53 #include "gromacs/mdrunutility/multisim.h"
  54 #include "gromacs/utility/gmxmpi.h"
  55
  56 struct gmx_multisim_t;
  57
  58 namespace gmx
  59 {
  60
  61 template<typename T>
  62 class ArrayRef;
  63
  64 /*! \cond libapi
  65  * \libinternal
  66  * \brief Simulation environment and configuration.
  67  *
  68  * SimulationContext allows a simulation module (\relates gmx::mdrun) to retrieve
  69  * runtime parameters and resources from client code. The client retains ownership
  70  * of the context and its resources, with exceptions as noted.
  71  *
  72  * A client must share ownership of some resources and transfer ownership of
  73  * other resources to create or configure the context. The simulation may in
  74  * turn consume or borrow some resources from the context. This is a new
  75  * framework that will evolve in the contexts of
  76  * https://gitlab.com/gromacs/gromacs/-/issues/2375
  77  * https://gitlab.com/gromacs/gromacs/-/issues/2587
  78  *
  79  * The public interface of SimulationContext is not yet well-specified.
  80  * Client code can create an instance with gmx::createSimulationContext()
  81  *
  82  * \warning The SimulationContext does not yet encapsulate the resources allocated
  83  * to the simulator. See https://gitlab.com/gromacs/gromacs/-/issues/3650
  84  *
  85  * \warning On thread-MPI code paths, the SimulationContext *does not* represent
  86  * constraints on the number of simulation ranks, nor does it represent initialized
  87  * communication resources.
  88  *
  89  * \todo Clarify and strengthen the invariant represented by SimulationContext.
  90  *
  91  * \todo This class should also handle aspects of simulation
  92  * environment such as working directory and environment variables.
  93  *
  94  * \ingroup module_mdrun
  95  * \inlibraryapi
  96  *
  97  * \internal
  98  * This is a minimal placeholder for a more complete implementation.
  99  * Interfaces for different API levels are not yet final.
 100  * \todo Impose sensible access restrictions.
 101  * Either the SimulationContext should be passed to the Mdrunner as logically constant or
 102  * a separate handle class can provide access to resources that have been
 103  * allocated by (negotiated with) the client for the current simulation
 104  * (or simulation segment).
 105  * Non-const accessors to shared resources need to be associated with update
 106  * signals that the simulation components (modules and runner) can subscribe to.
 107  *
 108  * Also reference https://gitlab.com/gromacs/gromacs/-/issues/2587
 109  */
 110 class SimulationContext final
 111 {
 112 public:
 113     /*!
 114      * \brief Object must be initialized with non-default constructor.
 115      */
 116     SimulationContext() = delete;
 117     /*!
 118      * \brief Initialize from borrowed communicator.
 119      *
 120      * \param communicator Communicator for all collaborating simulation processes.
 121      * \param multiSimDirectoryNames  Names of any directories used with -multidir
 122      *
 123      * Caller is responsible for keeping *communicator* valid for the life of SimulationContext,
 124      * and then freeing the communicator, if appropriate.
 125      *
 126      * With an external MPI library (non-thread-MPI chosen when configuring with CMake),
 127      * the client promises that MPI has been initialized (such as by calling gmx::init()).
 128      * This communicator is "borrowed" (not duplicated) from the caller.
 129      * Additional communicators may be split from the provided communicator
 130      * during the life of the SimulationContext or its consumers.
 131      *
 132      * With thread-MPI, *communicator* must be MPI_COMM_NULL.
 133      * The communicator is set up later
 134      * during the process of spawning the threads that will be the MPI
 135      * ranks. (Multi-simulation is not supported with thread-MPI.)
 136      *
 137      * \todo Refine distribution of environment management.
 138      *       There should be a context level at which only the current simulator directory matters,
 139      *       and a level above which encapsulates multisim details in a specialized type.
 140      */
 141     explicit SimulationContext(MPI_Comm communicator, ArrayRef<const std::string> multiSimDirectoryNames);
 142
 143     /*!
 144      * \brief MPI resources for the entire simulation context.
 145      *
 146      * With an external MPI library (non-thread-MPI chosen when configuring with CMake),
 147      * gmx::init() has called MPI_Init and the provided communicator is valid to use.
 148      * The communicator is "borrowed" (not duplicated) from the caller.
 149      *
 150      * With thread-MPI, the communicator is set up later
 151      * during the process of spawning the threads that will be the MPI
 152      * ranks.
 153      */
 154     MPI_Comm libraryWorldCommunicator_ = MPI_COMM_NULL;
 155
 156     /*!
 157      * \brief MPI communicator object for this simulation.
 158      *
 159      * With an external MPI library (non-thread-MPI chosen when configuring with CMake),
 160      * gmx::init() has called MPI_Init and the provided communicator is valid to use.
 161      * The communicator is "borrowed" (not duplicated) from the caller.
 162      *
 163      * With thread-MPI, the communicator is set up later
 164      * during the process of spawning the threads that will be the MPI
 165      * ranks.
 166      */
 167     MPI_Comm simulationCommunicator_ = MPI_COMM_NULL;
 168
 169     /*!
 170      * \brief Multi-sim handler (if required by e.g. gmx mdrun
 171      * -multidir; only supported with real MPI)
 172      *
 173      * If a multi-simulation is in use, then its
 174      * communicator(s) are found in multiSimulation_. This
 175      * communicator is that of all ranks from all simulations, and
 176      * will later be split into one for each simulation.
 177      * TODO: Perhaps (for simplicity) communicator splitting
 178      *       can be undertaken during multi-sim setup (acquisition of the multisim resource).
 179      *
 180      * Multi-simulation is not supported with thread-MPI.
 181      */
 182     std::unique_ptr<gmx_multisim_t> multiSimulation_;
 183 };
 184 //! \endcond
 185
 186 } // end namespace gmx
 187
 188 #endif // GROMACS_SIMULATIONCONTEXT_H