src/gromacs/mdlib/gmx_omp_nthreads.h

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2016,2017,2018, 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
  36 #ifndef GMX_OMP_NTHREADS_H
  37 #define GMX_OMP_NTHREADS_H
  38
  39 #include <stdio.h>
  40
  41 #include "gromacs/utility/basedefinitions.h"
  42
  43 struct t_commrec;
  44
  45 namespace gmx
  46 {
  47 class MDLogger;
  48 }
  49
  50 /** Enum values corresponding to multithreaded algorithmic modules. */
  51 typedef enum module_nth
  52 {
  53     /* Default is meant to be used in OMP regions outside the named
  54      * algorithmic modules listed below. */
  55     emntDefault, emntDomdec, emntPairsearch, emntNonbonded,
  56     emntBonded, emntPME,  emntUpdate, emntVSITE, emntLINCS, emntSETTLE,
  57     emntNR
  58 } module_nth_t;
  59
  60 /*! \brief
  61  * Initializes the per-module thread count.
  62  *
  63  * It is compatible with tMPI, thread-safety is ensured (for the features
  64  * available with tMPI).
  65  * This function should caled only once during the initialization of mdrun. */
  66 void gmx_omp_nthreads_init(const gmx::MDLogger &fplog, t_commrec *cr,
  67                            int nthreads_hw_avail,
  68                            int numRanksOnThisNode,
  69                            int omp_nthreads_req,
  70                            int omp_nthreads_pme_req,
  71                            gmx_bool bCurrNodePMEOnly,
  72                            gmx_bool bFullOmpSupport);
  73
  74 /*! \brief
  75  * Returns the number of threads to be used in the given module \p mod. */
  76 int gmx_omp_nthreads_get(int mod);
  77
  78 /*! \brief
  79  * Returns the number of threads to be used in the given module \p mod for simple rvec operations.
  80  *
  81  * When the, potentially, parallel task only consists of a loop of clear_rvec
  82  * or rvec_inc for nrvec elements, the OpenMP overhead might be higher than
  83  * the reduction in computional cost due to parallelization. This routine
  84  * returns 1 when the overhead is expected to be higher than the gain.
  85  */
  86 static inline int gmx_omp_nthreads_get_simple_rvec_task(int mod, int nrvec)
  87 {
  88     /* There can be a relatively large overhead to an OpenMP parallel for loop.
  89      * This overhead increases, slowly, with the numbe of threads used.
  90      * The computational gain goes as 1/#threads. The two effects combined
  91      * lead to a cross-over point for a (non-)parallel loop at loop count
  92      * that is not strongly dependent on the thread count.
  93      * Note that a (non-)parallel loop can have benefit later in the code
  94      * due to generating more cache hits, depending on how the next lask
  95      * that accesses the same data is (not) parallelized over threads.
  96      *
  97      * A value of 2000 is the switch-over point for Haswell without
  98      * hyper-threading. With hyper-threading it is about a factor 1.5 higher.
  99      */
 100     const int nrvec_omp = 2000;
 101
 102     if (nrvec < nrvec_omp)
 103     {
 104         return 1;
 105     }
 106     else
 107     {
 108         return gmx_omp_nthreads_get(mod);
 109     }
 110 }
 111
 112 /*! \brief Sets the number of threads to be used in module.
 113  *
 114  * Intended for use in testing. */
 115 void gmx_omp_nthreads_set(int mod, int nthreads);
 116
 117 /*! \brief
 118  * Read the OMP_NUM_THREADS env. var. and check against the value set on the
 119  * command line. */
 120 void gmx_omp_nthreads_read_env(const gmx::MDLogger &mdlog,
 121                                int                 *nthreads_omp);
 122
 123 #endif