src/gromacs/hardware/prepare_detection.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2016 by the GROMACS development team.
   5  * Copyright (c) 2017,2018,2019,2020, by the GROMACS development team, led by
   6  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   7  * and including many others, as listed in the AUTHORS file in the
   8  * top-level source directory and at http://www.gromacs.org.
   9  *
  10  * GROMACS is free software; you can redistribute it and/or
  11  * modify it under the terms of the GNU Lesser General Public License
  12  * as published by the Free Software Foundation; either version 2.1
  13  * of the License, or (at your option) any later version.
  14  *
  15  * GROMACS is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18  * Lesser General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public
  21  * License along with GROMACS; if not, see
  22  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  23  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  24  *
  25  * If you want to redistribute modifications to GROMACS, please
  26  * consider that scientific software is very special. Version
  27  * control is crucial - bugs must be traceable. We will be happy to
  28  * consider code for inclusion in the official distribution, but
  29  * derived work must not be called official GROMACS. Details are found
  30  * in the README & COPYING files - if they are missing, get the
  31  * official version at http://www.gromacs.org.
  32  *
  33  * To help us fund GROMACS development, we humbly ask that you cite
  34  * the research papers on the package. Check out http://www.gromacs.org.
  35  */
  36 /*! \internal \file
  37  * \brief Defines routine for activating potentially deactivated cores
  38  * so they can be detected.
  39  *
  40  * The use of std::thread makes for brittle interaction with std
  41  * library headers. Its caller also handles GPU detection and
  42  * allocation of device-specific data structures. This is more
  43  * manageable when separated into two distinct translation units.
  44  *
  45  * \author Erik Lindahl <erik.lindahl@scilifelab.se>
  46  * \author Mark Abraham <mark.j.abraham@gmail.com>
  47  * \ingroup module_hardware
  48  */
  49 #include "gmxpre.h"
  50
  51 #include "prepare_detection.h"
  52
  53 #include "config.h"
  54
  55 #include <cstdio>
  56
  57 #include <chrono>
  58 #include <thread>
  59 #include <vector>
  60
  61 #include "architecture.h"
  62
  63 #ifdef HAVE_UNISTD_H
  64 #    include <unistd.h> // sysconf()
  65 #endif
  66
  67 namespace gmx
  68 {
  69
  70 /*! \brief Utility that does dummy computing for max 2 seconds to spin up cores
  71  *
  72  *  This routine will check the number of cores configured and online
  73  *  (using sysconf), and the spins doing dummy compute operations for up to
  74  *  2 seconds, or until all cores have come online. This can be used prior to
  75  *  hardware detection for platforms that take unused processors offline.
  76  *
  77  *  This routine will not throw exceptions. In principle it should be
  78  *  declared noexcept, but at least icc 19.1 and 21-beta08 with the
  79  *  libstdc++-7.5 has difficulty implementing a std::vector of
  80  *  std::thread started with this function when declared noexcept. It
  81  *  is a known compiler bug that should be fixed after 19.1.
  82  *  Fortunately, this function is not performance sensitive,
  83  *  and only runs on platforms other than x86 and POWER (ie ARM),
  84  *  so the possible overhead introduced by omitting noexcept is not
  85  *  important.
  86  */
  87 static void spinUpCore()
  88 {
  89 #if defined(HAVE_SYSCONF) && defined(_SC_NPROCESSORS_CONF) && defined(_SC_NPROCESSORS_ONLN)
  90     float dummy           = 0.1;
  91     int   countConfigured = sysconf(_SC_NPROCESSORS_CONF);    // noexcept
  92     auto  start           = std::chrono::steady_clock::now(); // noexcept
  93
  94     while (sysconf(_SC_NPROCESSORS_ONLN) < countConfigured
  95            && std::chrono::steady_clock::now() - start < std::chrono::seconds(2))
  96     {
  97         for (int i = 1; i < 10000; i++)
  98         {
  99             dummy /= i;
 100         }
 101     }
 102
 103     if (dummy < 0)
 104     {
 105         printf("This cannot happen, but prevents loop from being optimized away.");
 106     }
 107 #endif
 108 }
 109
 110 void hardwareTopologyPrepareDetection()
 111 {
 112 #if defined(HAVE_SYSCONF) && defined(_SC_NPROCESSORS_CONF) \
 113         && (defined(THREAD_PTHREADS) || defined(THREAD_WINDOWS))
 114
 115     // Modify this conditional when/if x86 or PowerPC starts to sleep some cores
 116     if (c_architecture != Architecture::X86 && c_architecture != Architecture::PowerPC)
 117     {
 118         int                      countConfigured = sysconf(_SC_NPROCESSORS_CONF);
 119         std::vector<std::thread> workThreads(countConfigured);
 120
 121         for (auto& t : workThreads)
 122         {
 123             t = std::thread(spinUpCore);
 124         }
 125
 126         for (auto& t : workThreads)
 127         {
 128             t.join();
 129         }
 130     }
 131 #endif
 132 }
 133
 134 } // namespace gmx