2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2004, The GROMACS development team.
6 * Copyright (c) 2011-2019,2020,2021, 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.
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.
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.
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.
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.
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.
39 * \brief Implements the MD runner routine calling all integrators.
41 * \author David van der Spoel <david.vanderspoel@icm.uu.se>
42 * \ingroup module_mdrun
59 #include "gromacs/commandline/filenm.h"
60 #include "gromacs/domdec/builder.h"
61 #include "gromacs/domdec/domdec.h"
62 #include "gromacs/domdec/domdec_struct.h"
63 #include "gromacs/domdec/gpuhaloexchange.h"
64 #include "gromacs/domdec/localatomsetmanager.h"
65 #include "gromacs/domdec/makebondedlinks.h"
66 #include "gromacs/domdec/partition.h"
67 #include "gromacs/domdec/reversetopology.h"
68 #include "gromacs/ewald/ewald_utils.h"
69 #include "gromacs/ewald/pme.h"
70 #include "gromacs/ewald/pme_gpu_program.h"
71 #include "gromacs/ewald/pme_only.h"
72 #include "gromacs/ewald/pme_pp_comm_gpu.h"
73 #include "gromacs/fileio/checkpoint.h"
74 #include "gromacs/fileio/gmxfio.h"
75 #include "gromacs/fileio/oenv.h"
76 #include "gromacs/fileio/tpxio.h"
77 #include "gromacs/gmxlib/network.h"
78 #include "gromacs/gmxlib/nrnb.h"
79 #include "gromacs/gpu_utils/device_stream_manager.h"
80 #include "gromacs/hardware/cpuinfo.h"
81 #include "gromacs/hardware/detecthardware.h"
82 #include "gromacs/hardware/device_management.h"
83 #include "gromacs/hardware/hardwaretopology.h"
84 #include "gromacs/hardware/printhardware.h"
85 #include "gromacs/imd/imd.h"
86 #include "gromacs/listed_forces/disre.h"
87 #include "gromacs/listed_forces/listed_forces_gpu.h"
88 #include "gromacs/listed_forces/listed_forces.h"
89 #include "gromacs/listed_forces/orires.h"
90 #include "gromacs/math/functions.h"
91 #include "gromacs/math/utilities.h"
92 #include "gromacs/math/vec.h"
93 #include "gromacs/mdlib/boxdeformation.h"
94 #include "gromacs/mdlib/broadcaststructs.h"
95 #include "gromacs/mdlib/calc_verletbuf.h"
96 #include "gromacs/mdlib/dispersioncorrection.h"
97 #include "gromacs/mdlib/enerdata_utils.h"
98 #include "gromacs/mdlib/force.h"
99 #include "gromacs/mdlib/forcerec.h"
100 #include "gromacs/mdlib/gmx_omp_nthreads.h"
101 #include "gromacs/mdlib/gpuforcereduction.h"
102 #include "gromacs/mdlib/makeconstraints.h"
103 #include "gromacs/mdlib/md_support.h"
104 #include "gromacs/mdlib/mdatoms.h"
105 #include "gromacs/mdlib/sighandler.h"
106 #include "gromacs/mdlib/stophandler.h"
107 #include "gromacs/mdlib/tgroup.h"
108 #include "gromacs/mdlib/updategroups.h"
109 #include "gromacs/mdlib/vsite.h"
110 #include "gromacs/mdrun/mdmodules.h"
111 #include "gromacs/mdrun/simulationcontext.h"
112 #include "gromacs/mdrun/simulationinput.h"
113 #include "gromacs/mdrun/simulationinputhandle.h"
114 #include "gromacs/mdrunutility/handlerestart.h"
115 #include "gromacs/mdrunutility/logging.h"
116 #include "gromacs/mdrunutility/multisim.h"
117 #include "gromacs/mdrunutility/printtime.h"
118 #include "gromacs/mdrunutility/threadaffinity.h"
119 #include "gromacs/mdtypes/checkpointdata.h"
120 #include "gromacs/mdtypes/commrec.h"
121 #include "gromacs/mdtypes/enerdata.h"
122 #include "gromacs/mdtypes/fcdata.h"
123 #include "gromacs/mdtypes/forcerec.h"
124 #include "gromacs/mdtypes/group.h"
125 #include "gromacs/mdtypes/inputrec.h"
126 #include "gromacs/mdtypes/interaction_const.h"
127 #include "gromacs/mdtypes/md_enums.h"
128 #include "gromacs/mdtypes/mdatom.h"
129 #include "gromacs/mdtypes/mdrunoptions.h"
130 #include "gromacs/mdtypes/observableshistory.h"
131 #include "gromacs/mdtypes/simulation_workload.h"
132 #include "gromacs/mdtypes/state.h"
133 #include "gromacs/mdtypes/state_propagator_data_gpu.h"
134 #include "gromacs/modularsimulator/modularsimulator.h"
135 #include "gromacs/nbnxm/gpu_data_mgmt.h"
136 #include "gromacs/nbnxm/nbnxm.h"
137 #include "gromacs/nbnxm/pairlist_tuning.h"
138 #include "gromacs/pbcutil/pbc.h"
139 #include "gromacs/pulling/output.h"
140 #include "gromacs/pulling/pull.h"
141 #include "gromacs/pulling/pull_rotation.h"
142 #include "gromacs/restraint/manager.h"
143 #include "gromacs/restraint/restraintmdmodule.h"
144 #include "gromacs/restraint/restraintpotential.h"
145 #include "gromacs/swap/swapcoords.h"
146 #include "gromacs/taskassignment/decidegpuusage.h"
147 #include "gromacs/taskassignment/decidesimulationworkload.h"
148 #include "gromacs/taskassignment/resourcedivision.h"
149 #include "gromacs/taskassignment/taskassignment.h"
150 #include "gromacs/taskassignment/usergpuids.h"
151 #include "gromacs/timing/gpu_timing.h"
152 #include "gromacs/timing/wallcycle.h"
153 #include "gromacs/timing/wallcyclereporting.h"
154 #include "gromacs/topology/mtop_util.h"
155 #include "gromacs/trajectory/trajectoryframe.h"
156 #include "gromacs/utility/basenetwork.h"
157 #include "gromacs/utility/cstringutil.h"
158 #include "gromacs/utility/exceptions.h"
159 #include "gromacs/utility/fatalerror.h"
160 #include "gromacs/utility/filestream.h"
161 #include "gromacs/utility/gmxassert.h"
162 #include "gromacs/utility/gmxmpi.h"
163 #include "gromacs/utility/keyvaluetree.h"
164 #include "gromacs/utility/logger.h"
165 #include "gromacs/utility/loggerbuilder.h"
166 #include "gromacs/utility/mdmodulesnotifiers.h"
167 #include "gromacs/utility/physicalnodecommunicator.h"
168 #include "gromacs/utility/pleasecite.h"
169 #include "gromacs/utility/programcontext.h"
170 #include "gromacs/utility/smalloc.h"
171 #include "gromacs/utility/stringutil.h"
172 #include "gromacs/utility/mpiinfo.h"
174 #include "isimulator.h"
175 #include "membedholder.h"
176 #include "replicaexchange.h"
177 #include "simulatorbuilder.h"
183 /*! \brief Manage any development feature flag variables encountered
185 * The use of dev features indicated by environment variables is
186 * logged in order to ensure that runs with such features enabled can
187 * be identified from their log and standard output. Any cross
188 * dependencies are also checked, and if unsatisfied, a fatal error
191 * Note that some development features overrides are applied already here:
192 * the GPU communication flags are set to false in non-tMPI and non-CUDA builds.
194 * \param[in] mdlog Logger object.
195 * \param[in] useGpuForNonbonded True if the nonbonded task is offloaded in this run.
196 * \param[in] pmeRunMode The PME run mode for this run
197 * \returns The object populated with development feature flags.
199 static DevelopmentFeatureFlags manageDevelopmentFeatures(const gmx::MDLogger& mdlog,
200 const bool useGpuForNonbonded,
201 const PmeRunMode pmeRunMode)
203 DevelopmentFeatureFlags devFlags;
205 // Some builds of GCC 5 give false positive warnings that these
206 // getenv results are ignored when clearly they are used.
207 #pragma GCC diagnostic push
208 #pragma GCC diagnostic ignored "-Wunused-result"
210 devFlags.enableGpuBufferOps =
211 GMX_GPU_CUDA && useGpuForNonbonded && (getenv("GMX_USE_GPU_BUFFER_OPS") != nullptr);
212 devFlags.enableGpuHaloExchange = GMX_GPU_CUDA && getenv("GMX_GPU_DD_COMMS") != nullptr;
213 devFlags.forceGpuUpdateDefault = (getenv("GMX_FORCE_UPDATE_DEFAULT_GPU") != nullptr) || GMX_FAHCORE;
214 devFlags.enableGpuPmePPComm = GMX_GPU_CUDA && getenv("GMX_GPU_PME_PP_COMMS") != nullptr;
216 #pragma GCC diagnostic pop
218 // Direct GPU comm path is being used with CUDA_AWARE_MPI
219 // make sure underlying MPI implementation is CUDA-aware
220 if (!GMX_THREAD_MPI && (devFlags.enableGpuPmePPComm || devFlags.enableGpuHaloExchange))
222 const bool haveDetectedCudaAwareMpi =
223 (checkMpiCudaAwareSupport() == CudaAwareMpiStatus::Supported);
224 const bool forceCudaAwareMpi = (getenv("GMX_FORCE_CUDA_AWARE_MPI") != nullptr);
226 if (!haveDetectedCudaAwareMpi && forceCudaAwareMpi)
228 // CUDA-aware support not detected in MPI library but, user has forced it's use
229 GMX_LOG(mdlog.warning)
231 .appendTextFormatted(
232 "This run has forced use of 'CUDA-aware MPI'. "
233 "But, GROMACS cannot determine if underlying MPI "
234 "is CUDA-aware. GROMACS recommends use of latest openMPI version "
235 "for CUDA-aware support. "
236 "If you observe failures at runtime, try unsetting "
237 "GMX_FORCE_CUDA_AWARE_MPI environment variable.");
240 if (haveDetectedCudaAwareMpi || forceCudaAwareMpi)
242 devFlags.usingCudaAwareMpi = true;
243 GMX_LOG(mdlog.warning)
245 .appendTextFormatted(
246 "Using CUDA-aware MPI for 'GPU halo exchange' or 'GPU PME-PP "
247 "communications' feature.");
251 if (devFlags.enableGpuHaloExchange)
253 GMX_LOG(mdlog.warning)
255 .appendTextFormatted(
256 "GMX_GPU_DD_COMMS environment variable detected, but the 'GPU "
257 "halo exchange' feature will not be enabled as GROMACS couldn't "
258 "detect CUDA_aware support in underlying MPI implementation.");
259 devFlags.enableGpuHaloExchange = false;
261 if (devFlags.enableGpuPmePPComm)
263 GMX_LOG(mdlog.warning)
266 "GMX_GPU_PME_PP_COMMS environment variable detected, but the "
267 "'GPU PME-PP communications' feature will not be enabled as "
269 "detect CUDA_aware support in underlying MPI implementation.");
270 devFlags.enableGpuPmePPComm = false;
273 GMX_LOG(mdlog.warning)
275 .appendTextFormatted(
276 "GROMACS recommends use of latest OpenMPI version for CUDA-aware "
278 "If you are certain about CUDA-aware support in your MPI library, "
279 "you can force it's use by setting environment variable "
280 " GMX_FORCE_CUDA_AWARE_MPI.");
284 if (devFlags.enableGpuBufferOps)
286 GMX_LOG(mdlog.warning)
288 .appendTextFormatted(
289 "This run uses the 'GPU buffer ops' feature, enabled by the "
290 "GMX_USE_GPU_BUFFER_OPS environment variable.");
293 if (devFlags.forceGpuUpdateDefault)
295 GMX_LOG(mdlog.warning)
297 .appendTextFormatted(
298 "This run will default to '-update gpu' as requested by the "
299 "GMX_FORCE_UPDATE_DEFAULT_GPU environment variable. GPU update with domain "
300 "decomposition lacks substantial testing and should be used with caution.");
303 if (devFlags.enableGpuHaloExchange)
305 if (useGpuForNonbonded)
307 if (!devFlags.enableGpuBufferOps)
309 GMX_LOG(mdlog.warning)
311 .appendTextFormatted(
312 "Enabling GPU buffer operations required by GMX_GPU_DD_COMMS "
313 "(equivalent with GMX_USE_GPU_BUFFER_OPS=1).");
314 devFlags.enableGpuBufferOps = true;
316 GMX_LOG(mdlog.warning)
318 .appendTextFormatted(
319 "This run has requested the 'GPU halo exchange' feature, enabled by "
321 "GMX_GPU_DD_COMMS environment variable.");
325 GMX_LOG(mdlog.warning)
327 .appendTextFormatted(
328 "GMX_GPU_DD_COMMS environment variable detected, but the 'GPU "
329 "halo exchange' feature will not be enabled as nonbonded interactions "
330 "are not offloaded.");
331 devFlags.enableGpuHaloExchange = false;
335 if (devFlags.enableGpuPmePPComm)
337 if (pmeRunMode == PmeRunMode::GPU)
339 if (!devFlags.enableGpuBufferOps)
341 GMX_LOG(mdlog.warning)
343 .appendTextFormatted(
344 "Enabling GPU buffer operations required by GMX_GPU_PME_PP_COMMS "
345 "(equivalent with GMX_USE_GPU_BUFFER_OPS=1).");
346 devFlags.enableGpuBufferOps = true;
348 GMX_LOG(mdlog.warning)
350 .appendTextFormatted(
351 "This run uses the 'GPU PME-PP communications' feature, enabled "
352 "by the GMX_GPU_PME_PP_COMMS environment variable.");
356 std::string clarification;
357 if (pmeRunMode == PmeRunMode::Mixed)
360 "PME FFT and gather are not offloaded to the GPU (PME is running in mixed "
365 clarification = "PME is not offloaded to the GPU.";
367 GMX_LOG(mdlog.warning)
370 "GMX_GPU_PME_PP_COMMS environment variable detected, but the "
371 "'GPU PME-PP communications' feature was not enabled as "
373 devFlags.enableGpuPmePPComm = false;
380 /*! \brief Barrier for safe simultaneous thread access to mdrunner data
382 * Used to ensure that the master thread does not modify mdrunner during copy
383 * on the spawned threads. */
384 static void threadMpiMdrunnerAccessBarrier()
387 MPI_Barrier(MPI_COMM_WORLD);
391 Mdrunner Mdrunner::cloneOnSpawnedThread() const
393 auto newRunner = Mdrunner(std::make_unique<MDModules>());
395 // All runners in the same process share a restraint manager resource because it is
396 // part of the interface to the client code, which is associated only with the
397 // original thread. Handles to the same resources can be obtained by copy.
399 newRunner.restraintManager_ = std::make_unique<RestraintManager>(*restraintManager_);
402 // Copy members of master runner.
403 // \todo Replace with builder when Simulation context and/or runner phases are better defined.
404 // Ref https://gitlab.com/gromacs/gromacs/-/issues/2587 and https://gitlab.com/gromacs/gromacs/-/issues/2375
405 newRunner.hw_opt = hw_opt;
406 newRunner.filenames = filenames;
408 newRunner.hwinfo_ = hwinfo_;
409 newRunner.oenv = oenv;
410 newRunner.mdrunOptions = mdrunOptions;
411 newRunner.domdecOptions = domdecOptions;
412 newRunner.nbpu_opt = nbpu_opt;
413 newRunner.pme_opt = pme_opt;
414 newRunner.pme_fft_opt = pme_fft_opt;
415 newRunner.bonded_opt = bonded_opt;
416 newRunner.update_opt = update_opt;
417 newRunner.nstlist_cmdline = nstlist_cmdline;
418 newRunner.replExParams = replExParams;
419 newRunner.pforce = pforce;
420 // Give the spawned thread the newly created valid communicator
421 // for the simulation.
422 newRunner.libraryWorldCommunicator = MPI_COMM_WORLD;
423 newRunner.simulationCommunicator = MPI_COMM_WORLD;
425 newRunner.startingBehavior = startingBehavior;
426 newRunner.stopHandlerBuilder_ = std::make_unique<StopHandlerBuilder>(*stopHandlerBuilder_);
427 newRunner.inputHolder_ = inputHolder_;
429 threadMpiMdrunnerAccessBarrier();
434 /*! \brief The callback used for running on spawned threads.
436 * Obtains the pointer to the master mdrunner object from the one
437 * argument permitted to the thread-launch API call, copies it to make
438 * a new runner for this thread, reinitializes necessary data, and
439 * proceeds to the simulation. */
440 static void mdrunner_start_fn(const void* arg)
444 const auto* masterMdrunner = reinterpret_cast<const gmx::Mdrunner*>(arg);
445 /* copy the arg list to make sure that it's thread-local. This
446 doesn't copy pointed-to items, of course; fnm, cr and fplog
447 are reset in the call below, all others should be const. */
448 gmx::Mdrunner mdrunner = masterMdrunner->cloneOnSpawnedThread();
451 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
455 void Mdrunner::spawnThreads(int numThreadsToLaunch)
458 /* now spawn new threads that start mdrunner_start_fn(), while
459 the main thread returns. Thread affinity is handled later. */
460 if (tMPI_Init_fn(TRUE, numThreadsToLaunch, TMPI_AFFINITY_NONE, mdrunner_start_fn, static_cast<const void*>(this))
463 GMX_THROW(gmx::InternalError("Failed to spawn thread-MPI threads"));
466 // Give the master thread the newly created valid communicator for
468 libraryWorldCommunicator = MPI_COMM_WORLD;
469 simulationCommunicator = MPI_COMM_WORLD;
470 threadMpiMdrunnerAccessBarrier();
472 GMX_UNUSED_VALUE(numThreadsToLaunch);
473 GMX_UNUSED_VALUE(mdrunner_start_fn);
479 /*! \brief Initialize variables for Verlet scheme simulation */
480 static void prepare_verlet_scheme(FILE* fplog,
484 const gmx_mtop_t& mtop,
486 bool makeGpuPairList,
487 const gmx::CpuInfo& cpuinfo)
489 // We checked the cut-offs in grompp, but double-check here.
490 // We have PME+LJcutoff kernels for rcoulomb>rvdw.
491 if (EEL_PME_EWALD(ir->coulombtype) && ir->vdwtype == VanDerWaalsType::Cut)
493 GMX_RELEASE_ASSERT(ir->rcoulomb >= ir->rvdw,
494 "With Verlet lists and PME we should have rcoulomb>=rvdw");
498 GMX_RELEASE_ASSERT(ir->rcoulomb == ir->rvdw,
499 "With Verlet lists and no PME rcoulomb and rvdw should be identical");
501 /* For NVE simulations, we will retain the initial list buffer */
502 if (EI_DYNAMICS(ir->eI) && ir->verletbuf_tol > 0
503 && !(EI_MD(ir->eI) && ir->etc == TemperatureCoupling::No))
505 /* Update the Verlet buffer size for the current run setup */
507 /* Here we assume SIMD-enabled kernels are being used. But as currently
508 * calc_verlet_buffer_size gives the same results for 4x8 and 4x4
509 * and 4x2 gives a larger buffer than 4x4, this is ok.
511 ListSetupType listType =
512 (makeGpuPairList ? ListSetupType::Gpu : ListSetupType::CpuSimdWhenSupported);
513 VerletbufListSetup listSetup = verletbufGetSafeListSetup(listType);
515 const real rlist_new =
516 calcVerletBufferSize(mtop, det(box), *ir, ir->nstlist, ir->nstlist - 1, -1, listSetup);
518 if (rlist_new != ir->rlist)
520 if (fplog != nullptr)
523 "\nChanging rlist from %g to %g for non-bonded %dx%d atom kernels\n\n",
526 listSetup.cluster_size_i,
527 listSetup.cluster_size_j);
529 ir->rlist = rlist_new;
533 if (nstlist_cmdline > 0 && (!EI_DYNAMICS(ir->eI) || ir->verletbuf_tol <= 0))
536 "Can not set nstlist without %s",
537 !EI_DYNAMICS(ir->eI) ? "dynamics" : "verlet-buffer-tolerance");
540 if (EI_DYNAMICS(ir->eI))
542 /* Set or try nstlist values */
543 increaseNstlist(fplog, cr, ir, nstlist_cmdline, &mtop, box, makeGpuPairList, cpuinfo);
547 /*! \brief Override the nslist value in inputrec
549 * with value passed on the command line (if any)
551 static void override_nsteps_cmdline(const gmx::MDLogger& mdlog, int64_t nsteps_cmdline, t_inputrec* ir)
555 /* override with anything else than the default -2 */
556 if (nsteps_cmdline > -2)
558 char sbuf_steps[STEPSTRSIZE];
559 char sbuf_msg[STRLEN];
561 ir->nsteps = nsteps_cmdline;
562 if (EI_DYNAMICS(ir->eI) && nsteps_cmdline != -1)
565 "Overriding nsteps with value passed on the command line: %s steps, %.3g ps",
566 gmx_step_str(nsteps_cmdline, sbuf_steps),
567 fabs(nsteps_cmdline * ir->delta_t));
572 "Overriding nsteps with value passed on the command line: %s steps",
573 gmx_step_str(nsteps_cmdline, sbuf_steps));
576 GMX_LOG(mdlog.warning).asParagraph().appendText(sbuf_msg);
578 else if (nsteps_cmdline < -2)
580 gmx_fatal(FARGS, "Invalid nsteps value passed on the command line: %" PRId64, nsteps_cmdline);
582 /* Do nothing if nsteps_cmdline == -2 */
588 /*! \brief Return whether GPU acceleration of nonbondeds is supported with the given settings.
590 * If not, and if a warning may be issued, logs a warning about
591 * falling back to CPU code. With thread-MPI, only the first
592 * call to this function should have \c issueWarning true. */
593 static bool gpuAccelerationOfNonbondedIsUseful(const MDLogger& mdlog, const t_inputrec& ir, bool issueWarning)
595 bool gpuIsUseful = true;
598 if (ir.opts.ngener - ir.nwall > 1)
600 /* The GPU code does not support more than one energy group.
601 * If the user requested GPUs explicitly, a fatal error is given later.
605 "Multiple energy groups is not implemented for GPUs, falling back to the CPU. "
606 "For better performance, run on the GPU without energy groups and then do "
607 "gmx mdrun -rerun option on the trajectory with an energy group .tpr file.";
613 warning = "TPI is not implemented for GPUs.";
616 if (!gpuIsUseful && issueWarning)
618 GMX_LOG(mdlog.warning).asParagraph().appendText(warning);
624 //! Initializes the logger for mdrun.
625 static gmx::LoggerOwner buildLogger(FILE* fplog, const bool isSimulationMasterRank)
627 gmx::LoggerBuilder builder;
628 if (fplog != nullptr)
630 builder.addTargetFile(gmx::MDLogger::LogLevel::Info, fplog);
632 if (isSimulationMasterRank)
634 builder.addTargetStream(gmx::MDLogger::LogLevel::Warning, &gmx::TextOutputFile::standardError());
636 return builder.build();
639 //! Make a TaskTarget from an mdrun argument string.
640 static TaskTarget findTaskTarget(const char* optionString)
642 TaskTarget returnValue = TaskTarget::Auto;
644 if (strncmp(optionString, "auto", 3) == 0)
646 returnValue = TaskTarget::Auto;
648 else if (strncmp(optionString, "cpu", 3) == 0)
650 returnValue = TaskTarget::Cpu;
652 else if (strncmp(optionString, "gpu", 3) == 0)
654 returnValue = TaskTarget::Gpu;
658 GMX_ASSERT(false, "Option string should have been checked for sanity already");
664 //! Finish run, aggregate data to print performance info.
665 static void finish_run(FILE* fplog,
666 const gmx::MDLogger& mdlog,
668 const t_inputrec& inputrec,
670 gmx_wallcycle* wcycle,
671 gmx_walltime_accounting_t walltime_accounting,
672 nonbonded_verlet_t* nbv,
673 const gmx_pme_t* pme,
677 double nbfs = 0, mflop = 0;
678 double elapsed_time, elapsed_time_over_all_ranks, elapsed_time_over_all_threads,
679 elapsed_time_over_all_threads_over_all_ranks;
680 /* Control whether it is valid to print a report. Only the
681 simulation master may print, but it should not do so if the run
682 terminated e.g. before a scheduled reset step. This is
683 complicated by the fact that PME ranks are unaware of the
684 reason why they were sent a pmerecvqxFINISH. To avoid
685 communication deadlocks, we always do the communication for the
686 report, even if we've decided not to write the report, because
687 how long it takes to finish the run is not important when we've
688 decided not to report on the simulation performance.
690 Further, we only report performance for dynamical integrators,
691 because those are the only ones for which we plan to
692 consider doing any optimizations. */
693 bool printReport = EI_DYNAMICS(inputrec.eI) && SIMMASTER(cr);
695 if (printReport && !walltime_accounting_get_valid_finish(walltime_accounting))
697 GMX_LOG(mdlog.warning)
699 .appendText("Simulation ended prematurely, no performance report will be written.");
704 std::unique_ptr<t_nrnb> nrnbTotalStorage;
707 nrnbTotalStorage = std::make_unique<t_nrnb>();
708 nrnb_tot = nrnbTotalStorage.get();
710 MPI_Allreduce(nrnb->n.data(), nrnb_tot->n.data(), eNRNB, MPI_DOUBLE, MPI_SUM, cr->mpi_comm_mysim);
718 elapsed_time = walltime_accounting_get_time_since_reset(walltime_accounting);
719 elapsed_time_over_all_threads =
720 walltime_accounting_get_time_since_reset_over_all_threads(walltime_accounting);
724 /* reduce elapsed_time over all MPI ranks in the current simulation */
725 MPI_Allreduce(&elapsed_time, &elapsed_time_over_all_ranks, 1, MPI_DOUBLE, MPI_SUM, cr->mpi_comm_mysim);
726 elapsed_time_over_all_ranks /= cr->nnodes;
727 /* Reduce elapsed_time_over_all_threads over all MPI ranks in the
728 * current simulation. */
729 MPI_Allreduce(&elapsed_time_over_all_threads,
730 &elapsed_time_over_all_threads_over_all_ranks,
739 elapsed_time_over_all_ranks = elapsed_time;
740 elapsed_time_over_all_threads_over_all_ranks = elapsed_time_over_all_threads;
745 print_flop(fplog, nrnb_tot, &nbfs, &mflop);
748 if (thisRankHasDuty(cr, DUTY_PP) && DOMAINDECOMP(cr))
750 print_dd_statistics(cr, inputrec, fplog);
753 /* TODO Move the responsibility for any scaling by thread counts
754 * to the code that handled the thread region, so that there's a
755 * mechanism to keep cycle counting working during the transition
756 * to task parallelism. */
757 int nthreads_pp = gmx_omp_nthreads_get(ModuleMultiThread::Nonbonded);
758 int nthreads_pme = gmx_omp_nthreads_get(ModuleMultiThread::Pme);
759 wallcycle_scale_by_num_threads(
760 wcycle, thisRankHasDuty(cr, DUTY_PME) && !thisRankHasDuty(cr, DUTY_PP), nthreads_pp, nthreads_pme);
761 auto cycle_sum(wallcycle_sum(cr, wcycle));
765 auto* nbnxn_gpu_timings =
766 (nbv != nullptr && nbv->useGpu()) ? Nbnxm::gpu_get_timings(nbv->gpu_nbv) : nullptr;
767 gmx_wallclock_gpu_pme_t pme_gpu_timings = {};
769 if (pme_gpu_task_enabled(pme))
771 pme_gpu_get_timings(pme, &pme_gpu_timings);
773 wallcycle_print(fplog,
779 elapsed_time_over_all_ranks,
785 if (EI_DYNAMICS(inputrec.eI))
787 delta_t = inputrec.delta_t;
793 elapsed_time_over_all_threads_over_all_ranks,
794 elapsed_time_over_all_ranks,
795 walltime_accounting_get_nsteps_done_since_reset(walltime_accounting),
803 elapsed_time_over_all_threads_over_all_ranks,
804 elapsed_time_over_all_ranks,
805 walltime_accounting_get_nsteps_done_since_reset(walltime_accounting),
813 int Mdrunner::mdrunner()
816 std::unique_ptr<t_forcerec> fr;
817 real ewaldcoeff_q = 0;
818 real ewaldcoeff_lj = 0;
819 int nChargePerturbed = -1, nTypePerturbed = 0;
820 gmx_walltime_accounting_t walltime_accounting = nullptr;
821 MembedHolder membedHolder(filenames.size(), filenames.data());
823 /* CAUTION: threads may be started later on in this function, so
824 cr doesn't reflect the final parallel state right now */
827 /* TODO: inputrec should tell us whether we use an algorithm, not a file option */
828 const bool doEssentialDynamics = opt2bSet("-ei", filenames.size(), filenames.data());
829 const bool doRerun = mdrunOptions.rerun;
831 // Handle task-assignment related user options.
832 EmulateGpuNonbonded emulateGpuNonbonded =
833 (getenv("GMX_EMULATE_GPU") != nullptr ? EmulateGpuNonbonded::Yes : EmulateGpuNonbonded::No);
835 std::vector<int> userGpuTaskAssignment;
838 userGpuTaskAssignment = parseUserTaskAssignmentString(hw_opt.userGpuTaskAssignment);
840 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
841 auto nonbondedTarget = findTaskTarget(nbpu_opt);
842 auto pmeTarget = findTaskTarget(pme_opt);
843 auto pmeFftTarget = findTaskTarget(pme_fft_opt);
844 auto bondedTarget = findTaskTarget(bonded_opt);
845 auto updateTarget = findTaskTarget(update_opt);
847 FILE* fplog = nullptr;
848 // If we are appending, we don't write log output because we need
849 // to check that the old log file matches what the checkpoint file
850 // expects. Otherwise, we should start to write log output now if
851 // there is a file ready for it.
852 if (logFileHandle != nullptr && startingBehavior != StartingBehavior::RestartWithAppending)
854 fplog = gmx_fio_getfp(logFileHandle);
856 const bool isSimulationMasterRank = findIsSimulationMasterRank(ms, simulationCommunicator);
857 gmx::LoggerOwner logOwner(buildLogger(fplog, isSimulationMasterRank));
858 gmx::MDLogger mdlog(logOwner.logger());
860 gmx_print_detected_hardware(fplog, isSimulationMasterRank && isMasterSim(ms), mdlog, hwinfo_);
862 std::vector<int> availableDevices =
863 makeListOfAvailableDevices(hwinfo_->deviceInfoList, hw_opt.devicesSelectedByUser);
864 const int numAvailableDevices = gmx::ssize(availableDevices);
866 // Print citation requests after all software/hardware printing
867 pleaseCiteGromacs(fplog);
869 // Note: legacy program logic relies on checking whether these pointers are assigned.
870 // Objects may or may not be allocated later.
871 std::unique_ptr<t_inputrec> inputrec;
872 std::unique_ptr<t_state> globalState;
874 auto partialDeserializedTpr = std::make_unique<PartialDeserializedTprFile>();
876 if (isSimulationMasterRank)
878 // Allocate objects to be initialized by later function calls.
879 /* Only the master rank has the global state */
880 globalState = std::make_unique<t_state>();
881 inputrec = std::make_unique<t_inputrec>();
883 /* Read (nearly) all data required for the simulation
884 * and keep the partly serialized tpr contents to send to other ranks later
886 applyGlobalSimulationState(
887 *inputHolder_.get(), partialDeserializedTpr.get(), globalState.get(), inputrec.get(), &mtop);
890 /* Check and update the hardware options for internal consistency */
891 checkAndUpdateHardwareOptions(
892 mdlog, &hw_opt, isSimulationMasterRank, domdecOptions.numPmeRanks, inputrec.get());
894 if (GMX_THREAD_MPI && isSimulationMasterRank)
896 bool useGpuForNonbonded = false;
897 bool useGpuForPme = false;
900 GMX_RELEASE_ASSERT(inputrec != nullptr, "Keep the compiler happy");
902 // If the user specified the number of ranks, then we must
903 // respect that, but in default mode, we need to allow for
904 // the number of GPUs to choose the number of ranks.
905 auto canUseGpuForNonbonded = buildSupportsNonbondedOnGpu(nullptr);
906 useGpuForNonbonded = decideWhetherToUseGpusForNonbondedWithThreadMpi(
908 numAvailableDevices > 0,
909 userGpuTaskAssignment,
911 canUseGpuForNonbonded,
912 gpuAccelerationOfNonbondedIsUseful(mdlog, *inputrec, GMX_THREAD_MPI),
913 hw_opt.nthreads_tmpi);
914 useGpuForPme = decideWhetherToUseGpusForPmeWithThreadMpi(useGpuForNonbonded,
917 userGpuTaskAssignment,
920 hw_opt.nthreads_tmpi,
921 domdecOptions.numPmeRanks);
923 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
925 /* Determine how many thread-MPI ranks to start.
927 * TODO Over-writing the user-supplied value here does
928 * prevent any possible subsequent checks from working
930 hw_opt.nthreads_tmpi = get_nthreads_mpi(hwinfo_,
938 membedHolder.doMembed());
940 // Now start the threads for thread MPI.
941 spawnThreads(hw_opt.nthreads_tmpi);
942 // The spawned threads enter mdrunner() and execution of
943 // master and spawned threads joins at the end of this block.
946 GMX_RELEASE_ASSERT(ms || simulationCommunicator != MPI_COMM_NULL,
947 "Must have valid communicator unless running a multi-simulation");
948 CommrecHandle crHandle = init_commrec(simulationCommunicator);
949 t_commrec* cr = crHandle.get();
950 GMX_RELEASE_ASSERT(cr != nullptr, "Must have valid commrec");
952 PhysicalNodeCommunicator physicalNodeComm(libraryWorldCommunicator, gmx_physicalnode_id_hash());
954 // If we detected the topology on this system, double-check that it makes sense
955 if (hwinfo_->hardwareTopology->isThisSystem())
957 hardwareTopologyDoubleCheckDetection(mdlog, *hwinfo_->hardwareTopology);
962 /* now broadcast everything to the non-master nodes/threads: */
963 if (!isSimulationMasterRank)
965 // Until now, only the master rank has a non-null pointer.
966 // On non-master ranks, allocate the object that will receive data in the following call.
967 inputrec = std::make_unique<t_inputrec>();
969 init_parallel(cr->mpiDefaultCommunicator,
973 partialDeserializedTpr.get());
975 GMX_RELEASE_ASSERT(inputrec != nullptr, "All ranks should have a valid inputrec now");
976 partialDeserializedTpr.reset(nullptr);
979 !inputrec->useConstantAcceleration,
980 "Linear acceleration has been removed in GROMACS 2022, and was broken for many years "
981 "before that. Use GROMACS 4.5 or earlier if you need this feature.");
983 // Now the number of ranks is known to all ranks, and each knows
984 // the inputrec read by the master rank. The ranks can now all run
985 // the task-deciding functions and will agree on the result
986 // without needing to communicate.
987 const bool useDomainDecomposition =
988 (PAR(cr) && !(EI_TPI(inputrec->eI) || inputrec->eI == IntegrationAlgorithm::NM));
990 // Note that these variables describe only their own node.
992 // Note that when bonded interactions run on a GPU they always run
993 // alongside a nonbonded task, so do not influence task assignment
994 // even though they affect the force calculation workload.
995 bool useGpuForNonbonded = false;
996 bool useGpuForPme = false;
997 bool useGpuForBonded = false;
998 bool useGpuForUpdate = false;
999 bool gpusWereDetected = hwinfo_->ngpu_compatible_tot > 0;
1002 // It's possible that there are different numbers of GPUs on
1003 // different nodes, which is the user's responsibility to
1004 // handle. If unsuitable, we will notice that during task
1006 auto canUseGpuForNonbonded = buildSupportsNonbondedOnGpu(nullptr);
1007 useGpuForNonbonded = decideWhetherToUseGpusForNonbonded(
1009 userGpuTaskAssignment,
1010 emulateGpuNonbonded,
1011 canUseGpuForNonbonded,
1012 gpuAccelerationOfNonbondedIsUseful(mdlog, *inputrec, !GMX_THREAD_MPI),
1014 useGpuForPme = decideWhetherToUseGpusForPme(useGpuForNonbonded,
1016 userGpuTaskAssignment,
1019 cr->sizeOfDefaultCommunicator,
1020 domdecOptions.numPmeRanks,
1022 useGpuForBonded = decideWhetherToUseGpusForBonded(
1023 useGpuForNonbonded, useGpuForPme, bondedTarget, *inputrec, mtop, domdecOptions.numPmeRanks, gpusWereDetected);
1025 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
1027 const PmeRunMode pmeRunMode = determinePmeRunMode(useGpuForPme, pmeFftTarget, *inputrec);
1029 // Initialize development feature flags that enabled by environment variable
1030 // and report those features that are enabled.
1031 const DevelopmentFeatureFlags devFlags =
1032 manageDevelopmentFeatures(mdlog, useGpuForNonbonded, pmeRunMode);
1034 const bool useModularSimulator = checkUseModularSimulator(false,
1041 doEssentialDynamics,
1042 membedHolder.doMembed());
1044 // Build restraints.
1045 // TODO: hide restraint implementation details from Mdrunner.
1046 // There is nothing unique about restraints at this point as far as the
1047 // Mdrunner is concerned. The Mdrunner should just be getting a sequence of
1048 // factory functions from the SimulationContext on which to call mdModules_->add().
1049 // TODO: capture all restraints into a single RestraintModule, passed to the runner builder.
1050 for (auto&& restraint : restraintManager_->getRestraints())
1052 auto module = RestraintMDModule::create(restraint, restraint->sites());
1053 mdModules_->add(std::move(module));
1056 // TODO: Error handling
1057 mdModules_->assignOptionsToModules(*inputrec->params, nullptr);
1058 // now that the MDModules know their options, they know which callbacks to sign up to
1059 mdModules_->subscribeToSimulationSetupNotifications();
1060 const auto& setupNotifier = mdModules_->notifiers().simulationSetupNotifier_;
1062 if (inputrec->internalParameters != nullptr)
1064 setupNotifier.notify(*inputrec->internalParameters);
1067 if (fplog != nullptr)
1069 pr_inputrec(fplog, 0, "Input Parameters", inputrec.get(), FALSE);
1070 fprintf(fplog, "\n");
1075 /* In rerun, set velocities to zero if present */
1076 if (doRerun && ((globalState->flags & enumValueToBitMask(StateEntry::V)) != 0))
1078 // rerun does not use velocities
1082 "Rerun trajectory contains velocities. Rerun does only evaluate "
1083 "potential energy and forces. The velocities will be ignored.");
1084 for (int i = 0; i < globalState->natoms; i++)
1086 clear_rvec(globalState->v[i]);
1088 globalState->flags &= ~enumValueToBitMask(StateEntry::V);
1091 /* now make sure the state is initialized and propagated */
1092 set_state_entries(globalState.get(), inputrec.get(), useModularSimulator);
1095 /* NM and TPI parallelize over force/energy calculations, not atoms,
1096 * so we need to initialize and broadcast the global state.
1098 if (inputrec->eI == IntegrationAlgorithm::NM || inputrec->eI == IntegrationAlgorithm::TPI)
1102 globalState = std::make_unique<t_state>();
1104 broadcastStateWithoutDynamics(
1105 cr->mpiDefaultCommunicator, DOMAINDECOMP(cr), PAR(cr), globalState.get());
1108 /* A parallel command line option consistency check that we can
1109 only do after any threads have started. */
1111 && (domdecOptions.numCells[XX] > 1 || domdecOptions.numCells[YY] > 1
1112 || domdecOptions.numCells[ZZ] > 1 || domdecOptions.numPmeRanks > 0))
1115 "The -dd or -npme option request a parallel simulation, "
1117 "but %s was compiled without threads or MPI enabled",
1118 output_env_get_program_display_name(oenv));
1119 #elif GMX_THREAD_MPI
1120 "but the number of MPI-threads (option -ntmpi) is not set or is 1");
1122 "but %s was not started through mpirun/mpiexec or only one rank was requested "
1123 "through mpirun/mpiexec",
1124 output_env_get_program_display_name(oenv));
1128 if (doRerun && (EI_ENERGY_MINIMIZATION(inputrec->eI) || IntegrationAlgorithm::NM == inputrec->eI))
1131 "The .mdp file specified an energy mininization or normal mode algorithm, and "
1132 "these are not compatible with mdrun -rerun");
1135 /* NMR restraints must be initialized before load_checkpoint,
1136 * since with time averaging the history is added to t_state.
1137 * For proper consistency check we therefore need to extend
1139 * So the PME-only nodes (if present) will also initialize
1140 * the distance restraints.
1143 /* This needs to be called before read_checkpoint to extend the state */
1144 t_disresdata* disresdata;
1145 snew(disresdata, 1);
1149 DisResRunMode::MDRun,
1150 MASTER(cr) ? DDRole::Master : DDRole::Agent,
1151 PAR(cr) ? NumRanks::Multiple : NumRanks::Single,
1156 replExParams.exchangeInterval > 0);
1158 std::unique_ptr<t_oriresdata> oriresData;
1159 if (gmx_mtop_ftype_count(mtop, F_ORIRES) > 0)
1161 oriresData = std::make_unique<t_oriresdata>(fplog, mtop, *inputrec, cr, ms, globalState.get());
1164 auto deform = prepareBoxDeformation(globalState != nullptr ? globalState->box : box,
1165 MASTER(cr) ? DDRole::Master : DDRole::Agent,
1166 PAR(cr) ? NumRanks::Multiple : NumRanks::Single,
1167 cr->mpi_comm_mygroup,
1171 /* We have to remember the generation's first step before reading checkpoint.
1172 This way, we can report to the F@H core both the generation's first step
1173 and the restored first step, thus making it able to distinguish between
1174 an interruption/resume and start of the n-th generation simulation.
1175 Having this information, the F@H core can correctly calculate and report
1178 int gen_first_step = 0;
1181 gen_first_step = inputrec->init_step;
1185 ObservablesHistory observablesHistory = {};
1187 auto modularSimulatorCheckpointData = std::make_unique<ReadCheckpointDataHolder>();
1188 if (startingBehavior != StartingBehavior::NewSimulation)
1190 /* Check if checkpoint file exists before doing continuation.
1191 * This way we can use identical input options for the first and subsequent runs...
1193 if (mdrunOptions.numStepsCommandline > -2)
1195 /* Temporarily set the number of steps to unlimited to avoid
1196 * triggering the nsteps check in load_checkpoint().
1197 * This hack will go away soon when the -nsteps option is removed.
1199 inputrec->nsteps = -1;
1202 // Finish applying initial simulation state information from external sources on all ranks.
1203 // Reconcile checkpoint file data with Mdrunner state established up to this point.
1204 applyLocalState(*inputHolder_.get(),
1207 domdecOptions.numCells,
1210 &observablesHistory,
1211 mdrunOptions.reproducible,
1212 mdModules_->notifiers(),
1213 modularSimulatorCheckpointData.get(),
1214 useModularSimulator);
1215 // TODO: (#3652) Synchronize filesystem state, SimulationInput contents, and program
1217 // on all code paths.
1218 // Write checkpoint or provide hook to update SimulationInput.
1219 // If there was a checkpoint file, SimulationInput contains more information
1220 // than if there wasn't. At this point, we have synchronized the in-memory
1221 // state with the filesystem state only for restarted simulations. We should
1222 // be calling applyLocalState unconditionally and expect that the completeness
1223 // of SimulationInput is not dependent on its creation method.
1225 if (startingBehavior == StartingBehavior::RestartWithAppending && logFileHandle)
1227 // Now we can start normal logging to the truncated log file.
1228 fplog = gmx_fio_getfp(logFileHandle);
1229 prepareLogAppending(fplog);
1230 logOwner = buildLogger(fplog, MASTER(cr));
1231 mdlog = logOwner.logger();
1238 fcRegisterSteps(inputrec->nsteps + inputrec->init_step, gen_first_step);
1242 if (mdrunOptions.numStepsCommandline > -2)
1247 "The -nsteps functionality is deprecated, and may be removed in a future "
1249 "Consider using gmx convert-tpr -nsteps or changing the appropriate .mdp "
1252 /* override nsteps with value set on the commandline */
1253 override_nsteps_cmdline(mdlog, mdrunOptions.numStepsCommandline, inputrec.get());
1255 if (isSimulationMasterRank)
1257 copy_mat(globalState->box, box);
1262 gmx_bcast(sizeof(box), box, cr->mpiDefaultCommunicator);
1265 if (inputrec->cutoff_scheme != CutoffScheme::Verlet)
1268 "This group-scheme .tpr file can no longer be run by mdrun. Please update to the "
1269 "Verlet scheme, or use an earlier version of GROMACS if necessary.");
1271 /* Update rlist and nstlist. */
1272 /* Note: prepare_verlet_scheme is calling increaseNstlist(...), which (while attempting to
1273 * increase rlist) tries to check if the newly chosen value fits with the DD scheme. As this is
1274 * run before any DD scheme is set up, this check is never executed. See #3334 for more details.
1276 prepare_verlet_scheme(fplog,
1282 useGpuForNonbonded || (emulateGpuNonbonded == EmulateGpuNonbonded::Yes),
1285 // We need to decide on update groups early, as this affects
1286 // inter-domain communication distances.
1287 auto updateGroupingsPerMoleculeType = makeUpdateGroupingsPerMoleculeType(mtop);
1288 const real maxUpdateGroupRadius = computeMaxUpdateGroupRadius(
1289 mtop, updateGroupingsPerMoleculeType, maxReferenceTemperature(*inputrec));
1290 const real cutoffMargin = std::sqrt(max_cutoff2(inputrec->pbcType, box)) - inputrec->rlist;
1291 UpdateGroups updateGroups = makeUpdateGroups(mdlog,
1292 std::move(updateGroupingsPerMoleculeType),
1293 maxUpdateGroupRadius,
1294 useDomainDecomposition,
1295 systemHasConstraintsOrVsites(mtop),
1298 // This builder is necessary while we have multi-part construction
1299 // of DD. Before DD is constructed, we use the existence of
1300 // the builder object to indicate that further construction of DD
1302 std::unique_ptr<DomainDecompositionBuilder> ddBuilder;
1303 if (useDomainDecomposition)
1305 ddBuilder = std::make_unique<DomainDecompositionBuilder>(
1312 mdModules_->notifiers(),
1314 updateGroups.updateGroupingPerMoleculeType(),
1315 updateGroups.useUpdateGroups(),
1316 updateGroups.maxUpdateGroupRadius(),
1317 positionsFromStatePointer(globalState.get()),
1323 /* PME, if used, is done on all nodes with 1D decomposition */
1324 cr->nnodes = cr->sizeOfDefaultCommunicator;
1325 cr->sim_nodeid = cr->rankInDefaultCommunicator;
1326 cr->nodeid = cr->rankInDefaultCommunicator;
1328 cr->duty = (DUTY_PP | DUTY_PME);
1330 if (inputrec->pbcType == PbcType::Screw)
1332 gmx_fatal(FARGS, "pbc=screw is only implemented with domain decomposition");
1336 // Produce the task assignment for this rank - done after DD is constructed
1337 GpuTaskAssignments gpuTaskAssignments = GpuTaskAssignmentsBuilder::build(
1339 userGpuTaskAssignment,
1341 simulationCommunicator,
1349 thisRankHasDuty(cr, DUTY_PP),
1350 // TODO cr->duty & DUTY_PME should imply that a PME
1351 // algorithm is active, but currently does not.
1352 EEL_PME(inputrec->coulombtype) && thisRankHasDuty(cr, DUTY_PME));
1354 // Get the device handles for the modules, nullptr when no task is assigned.
1356 DeviceInformation* deviceInfo = gpuTaskAssignments.initDevice(&deviceId);
1358 // timing enabling - TODO put this in gpu_utils (even though generally this is just option handling?)
1359 bool useTiming = true;
1363 /* WARNING: CUDA timings are incorrect with multiple streams.
1364 * This is the main reason why they are disabled by default.
1366 // TODO: Consider turning on by default when we can detect nr of streams.
1367 useTiming = (getenv("GMX_ENABLE_GPU_TIMING") != nullptr);
1369 else if (GMX_GPU_OPENCL)
1371 useTiming = (getenv("GMX_DISABLE_GPU_TIMING") == nullptr);
1374 // TODO Currently this is always built, yet DD partition code
1375 // checks if it is built before using it. Probably it should
1376 // become an MDModule that is made only when another module
1377 // requires it (e.g. pull, CompEl, density fitting), so that we
1378 // don't update the local atom sets unilaterally every step.
1379 LocalAtomSetManager atomSets;
1382 // TODO Pass the GPU streams to ddBuilder to use in buffer
1383 // transfers (e.g. halo exchange)
1384 cr->dd = ddBuilder->build(&atomSets);
1385 // The builder's job is done, so destruct it
1386 ddBuilder.reset(nullptr);
1387 // Note that local state still does not exist yet.
1390 // The GPU update is decided here because we need to know whether the constraints or
1391 // SETTLEs can span accross the domain borders (i.e. whether or not update groups are
1392 // defined). This is only known after DD is initialized, hence decision on using GPU
1393 // update is done so late.
1396 const bool haveFrozenAtoms = inputrecFrozenAtoms(inputrec.get());
1398 useGpuForUpdate = decideWhetherToUseGpuForUpdate(useDomainDecomposition,
1399 updateGroups.useUpdateGroups(),
1401 domdecOptions.numPmeRanks > 0,
1407 doEssentialDynamics,
1408 gmx_mtop_ftype_count(mtop, F_ORIRES) > 0,
1414 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
1416 const bool printHostName = (cr->nnodes > 1);
1417 gpuTaskAssignments.reportGpuUsage(mdlog, printHostName, useGpuForBonded, pmeRunMode, useGpuForUpdate);
1419 const bool disableNonbondedCalculation = (getenv("GMX_NO_NONBONDED") != nullptr);
1420 if (disableNonbondedCalculation)
1422 /* turn off non-bonded calculations */
1423 GMX_LOG(mdlog.warning)
1426 "Found environment variable GMX_NO_NONBONDED.\n"
1427 "Disabling nonbonded calculations.");
1430 MdrunScheduleWorkload runScheduleWork;
1432 bool useGpuDirectHalo = decideWhetherToUseGpuForHalo(devFlags,
1433 havePPDomainDecomposition(cr),
1435 useModularSimulator,
1437 EI_ENERGY_MINIMIZATION(inputrec->eI));
1439 // Also populates the simulation constant workload description.
1440 runScheduleWork.simulationWork = createSimulationWorkload(*inputrec,
1441 disableNonbondedCalculation,
1449 std::unique_ptr<DeviceStreamManager> deviceStreamManager = nullptr;
1451 if (deviceInfo != nullptr)
1453 if (DOMAINDECOMP(cr) && thisRankHasDuty(cr, DUTY_PP))
1455 dd_setup_dlb_resource_sharing(cr, deviceId);
1457 deviceStreamManager = std::make_unique<DeviceStreamManager>(
1458 *deviceInfo, havePPDomainDecomposition(cr), runScheduleWork.simulationWork, useTiming);
1461 // If the user chose a task assignment, give them some hints
1462 // where appropriate.
1463 if (!userGpuTaskAssignment.empty())
1465 gpuTaskAssignments.logPerformanceHints(mdlog, numAvailableDevices);
1470 /* After possible communicator splitting in make_dd_communicators.
1471 * we can set up the intra/inter node communication.
1473 gmx_setup_nodecomm(fplog, cr);
1479 GMX_LOG(mdlog.warning)
1481 .appendTextFormatted(
1482 "This is simulation %d out of %d running as a composite GROMACS\n"
1483 "multi-simulation job. Setup for this simulation:\n",
1484 ms->simulationIndex_,
1485 ms->numSimulations_);
1487 GMX_LOG(mdlog.warning)
1488 .appendTextFormatted("Using %d MPI %s\n",
1491 cr->nnodes == 1 ? "thread" : "threads"
1493 cr->nnodes == 1 ? "process" : "processes"
1499 // If mdrun -pin auto honors any affinity setting that already
1500 // exists. If so, it is nice to provide feedback about whether
1501 // that existing affinity setting was from OpenMP or something
1502 // else, so we run this code both before and after we initialize
1503 // the OpenMP support.
1504 gmx_check_thread_affinity_set(mdlog, &hw_opt, hwinfo_->nthreads_hw_avail, FALSE);
1505 /* Check and update the number of OpenMP threads requested */
1506 checkAndUpdateRequestedNumOpenmpThreads(
1507 &hw_opt, *hwinfo_, cr, ms, physicalNodeComm.size_, pmeRunMode, mtop, *inputrec);
1509 gmx_omp_nthreads_init(mdlog,
1511 hwinfo_->nthreads_hw_avail,
1512 physicalNodeComm.size_,
1513 hw_opt.nthreads_omp,
1514 hw_opt.nthreads_omp_pme,
1515 !thisRankHasDuty(cr, DUTY_PP));
1517 const bool bEnableFPE = gmxShouldEnableFPExceptions();
1518 // FIXME - reconcile with gmx_feenableexcept() call from CommandLineModuleManager::run()
1521 gmx_feenableexcept();
1524 /* Now that we know the setup is consistent, check for efficiency */
1525 check_resource_division_efficiency(
1526 hwinfo_, gpuTaskAssignments.thisRankHasAnyGpuTask(), mdrunOptions.ntompOptionIsSet, cr, mdlog);
1528 /* getting number of PP/PME threads on this MPI / tMPI rank.
1529 PME: env variable should be read only on one node to make sure it is
1530 identical everywhere;
1532 const int numThreadsOnThisRank = thisRankHasDuty(cr, DUTY_PP)
1533 ? gmx_omp_nthreads_get(ModuleMultiThread::Nonbonded)
1534 : gmx_omp_nthreads_get(ModuleMultiThread::Pme);
1535 checkHardwareOversubscription(
1536 numThreadsOnThisRank, cr->nodeid, *hwinfo_->hardwareTopology, physicalNodeComm, mdlog);
1538 // Enable Peer access between GPUs where available
1539 // Only for DD, only master PP rank needs to perform setup, and only if thread MPI plus
1540 // any of the GPU communication features are active.
1541 if (DOMAINDECOMP(cr) && MASTER(cr) && thisRankHasDuty(cr, DUTY_PP) && GMX_THREAD_MPI
1542 && (runScheduleWork.simulationWork.useGpuHaloExchange
1543 || runScheduleWork.simulationWork.useGpuPmePpCommunication))
1545 setupGpuDevicePeerAccess(gpuTaskAssignments.deviceIdsAssigned(), mdlog);
1548 if (hw_opt.threadAffinity != ThreadAffinity::Off)
1550 /* Before setting affinity, check whether the affinity has changed
1551 * - which indicates that probably the OpenMP library has changed it
1552 * since we first checked).
1554 gmx_check_thread_affinity_set(mdlog, &hw_opt, hwinfo_->nthreads_hw_avail, TRUE);
1556 int numThreadsOnThisNode, intraNodeThreadOffset;
1557 analyzeThreadsOnThisNode(
1558 physicalNodeComm, numThreadsOnThisRank, &numThreadsOnThisNode, &intraNodeThreadOffset);
1560 /* Set the CPU affinity */
1561 gmx_set_thread_affinity(mdlog,
1564 *hwinfo_->hardwareTopology,
1565 numThreadsOnThisRank,
1566 numThreadsOnThisNode,
1567 intraNodeThreadOffset,
1571 if (mdrunOptions.timingOptions.resetStep > -1)
1576 "The -resetstep functionality is deprecated, and may be removed in a "
1579 std::unique_ptr<gmx_wallcycle> wcycle =
1580 wallcycle_init(fplog, mdrunOptions.timingOptions.resetStep, cr);
1584 /* Master synchronizes its value of reset_counters with all nodes
1585 * including PME only nodes */
1586 int64_t reset_counters = wcycle_get_reset_counters(wcycle.get());
1587 gmx_bcast(sizeof(reset_counters), &reset_counters, cr->mpi_comm_mysim);
1588 wcycle_set_reset_counters(wcycle.get(), reset_counters);
1591 // Membrane embedding must be initialized before we call init_forcerec()
1592 membedHolder.initializeMembed(fplog,
1599 &mdrunOptions.checkpointOptions.period);
1601 const bool thisRankHasPmeGpuTask = gpuTaskAssignments.thisRankHasPmeGpuTask();
1602 std::unique_ptr<MDAtoms> mdAtoms;
1603 std::unique_ptr<VirtualSitesHandler> vsite;
1604 std::unique_ptr<ListedForcesGpu> listedForcesGpu;
1607 if (thisRankHasDuty(cr, DUTY_PP))
1609 setupNotifier.notify(*cr);
1610 setupNotifier.notify(&atomSets);
1611 setupNotifier.notify(mtop);
1612 setupNotifier.notify(inputrec->pbcType);
1613 setupNotifier.notify(SimulationTimeStep{ inputrec->delta_t });
1614 /* Initiate forcerecord */
1615 fr = std::make_unique<t_forcerec>();
1616 fr->forceProviders = mdModules_->initForceProviders();
1617 init_forcerec(fplog,
1624 opt2fn("-table", filenames.size(), filenames.data()),
1625 opt2fn("-tablep", filenames.size(), filenames.data()),
1626 opt2fns("-tableb", filenames.size(), filenames.data()),
1628 // Dirty hack, for fixing disres and orires should be made mdmodules
1629 fr->fcdata->disres = disresdata;
1630 fr->fcdata->orires.swap(oriresData);
1632 // Save a handle to device stream manager to use elsewhere in the code
1633 // TODO: Forcerec is not a correct place to store it.
1634 fr->deviceStreamManager = deviceStreamManager.get();
1636 if (runScheduleWork.simulationWork.useGpuPmePpCommunication && !thisRankHasDuty(cr, DUTY_PME))
1639 deviceStreamManager != nullptr,
1640 "GPU device stream manager should be valid in order to use PME-PP direct "
1643 deviceStreamManager->streamIsValid(DeviceStreamType::PmePpTransfer),
1644 "GPU PP-PME stream should be valid in order to use GPU PME-PP direct "
1646 fr->pmePpCommGpu = std::make_unique<gmx::PmePpCommGpu>(
1649 deviceStreamManager->context(),
1650 deviceStreamManager->stream(DeviceStreamType::PmePpTransfer));
1653 fr->nbv = Nbnxm::init_nb_verlet(mdlog,
1658 runScheduleWork.simulationWork.useGpuNonbonded,
1659 deviceStreamManager.get(),
1663 // TODO: Move the logic below to a GPU bonded builder
1664 if (runScheduleWork.simulationWork.useGpuBonded)
1666 GMX_RELEASE_ASSERT(deviceStreamManager != nullptr,
1667 "GPU device stream manager should be valid in order to use GPU "
1668 "version of bonded forces.");
1669 listedForcesGpu = std::make_unique<ListedForcesGpu>(
1671 fr->ic->epsfac * fr->fudgeQQ,
1672 deviceStreamManager->context(),
1673 deviceStreamManager->bondedStream(havePPDomainDecomposition(cr)),
1675 fr->listedForcesGpu = listedForcesGpu.get();
1678 /* Initialize the mdAtoms structure.
1679 * mdAtoms is not filled with atom data,
1680 * as this can not be done now with domain decomposition.
1682 mdAtoms = makeMDAtoms(fplog, mtop, *inputrec, thisRankHasPmeGpuTask);
1683 if (globalState && thisRankHasPmeGpuTask)
1685 // The pinning of coordinates in the global state object works, because we only use
1686 // PME on GPU without DD or on a separate PME rank, and because the local state pointer
1687 // points to the global state object without DD.
1688 // FIXME: MD and EM separately set up the local state - this should happen in the same
1689 // function, which should also perform the pinning.
1690 changePinningPolicy(&globalState->x, pme_get_pinning_policy());
1693 /* Initialize the virtual site communication */
1694 vsite = makeVirtualSitesHandler(
1695 mtop, cr, fr->pbcType, updateGroups.updateGroupingPerMoleculeType());
1697 calc_shifts(box, fr->shift_vec);
1699 /* With periodic molecules the charge groups should be whole at start up
1700 * and the virtual sites should not be far from their proper positions.
1702 if (!inputrec->bContinuation && MASTER(cr)
1703 && !(inputrec->pbcType != PbcType::No && inputrec->bPeriodicMols))
1705 /* Make molecules whole at start of run */
1706 if (fr->pbcType != PbcType::No)
1708 do_pbc_first_mtop(fplog, inputrec->pbcType, box, &mtop, globalState->x.rvec_array());
1712 /* Correct initial vsite positions are required
1713 * for the initial distribution in the domain decomposition
1714 * and for the initial shell prediction.
1716 constructVirtualSitesGlobal(mtop, globalState->x);
1719 // Make the DD reverse topology, now that any vsites that are present are available
1720 if (DOMAINDECOMP(cr))
1722 dd_make_reverse_top(fplog, cr->dd, mtop, vsite.get(), *inputrec, domdecOptions.ddBondedChecking);
1725 if (EEL_PME(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype))
1727 ewaldcoeff_q = fr->ic->ewaldcoeff_q;
1728 ewaldcoeff_lj = fr->ic->ewaldcoeff_lj;
1733 /* This is a PME only node */
1735 GMX_ASSERT(globalState == nullptr,
1736 "We don't need the state on a PME only rank and expect it to be unitialized");
1738 ewaldcoeff_q = calc_ewaldcoeff_q(inputrec->rcoulomb, inputrec->ewald_rtol);
1739 ewaldcoeff_lj = calc_ewaldcoeff_lj(inputrec->rvdw, inputrec->ewald_rtol_lj);
1742 gmx_pme_t* sepPmeData = nullptr;
1743 // This reference hides the fact that PME data is owned by runner on PME-only ranks and by forcerec on other ranks
1744 GMX_ASSERT(thisRankHasDuty(cr, DUTY_PP) == (fr != nullptr),
1745 "Double-checking that only PME-only ranks have no forcerec");
1746 gmx_pme_t*& pmedata = fr ? fr->pmedata : sepPmeData;
1748 // TODO should live in ewald module once its testing is improved
1750 // Later, this program could contain kernels that might be later
1751 // re-used as auto-tuning progresses, or subsequent simulations
1753 PmeGpuProgramStorage pmeGpuProgram;
1754 if (thisRankHasPmeGpuTask)
1757 (deviceStreamManager != nullptr),
1758 "GPU device stream manager should be initialized in order to use GPU for PME.");
1759 GMX_RELEASE_ASSERT((deviceInfo != nullptr),
1760 "GPU device should be initialized in order to use GPU for PME.");
1761 pmeGpuProgram = buildPmeGpuProgram(deviceStreamManager->context());
1764 /* Initiate PME if necessary,
1765 * either on all nodes or on dedicated PME nodes only. */
1766 if (EEL_PME(inputrec->coulombtype) || EVDW_PME(inputrec->vdwtype))
1768 if (mdAtoms && mdAtoms->mdatoms())
1770 nChargePerturbed = mdAtoms->mdatoms()->nChargePerturbed;
1771 if (EVDW_PME(inputrec->vdwtype))
1773 nTypePerturbed = mdAtoms->mdatoms()->nTypePerturbed;
1776 if (cr->npmenodes > 0)
1778 /* The PME only nodes need to know nChargePerturbed(FEP on Q) and nTypePerturbed(FEP on LJ)*/
1779 gmx_bcast(sizeof(nChargePerturbed), &nChargePerturbed, cr->mpi_comm_mysim);
1780 gmx_bcast(sizeof(nTypePerturbed), &nTypePerturbed, cr->mpi_comm_mysim);
1783 if (thisRankHasDuty(cr, DUTY_PME))
1787 // TODO: This should be in the builder.
1788 GMX_RELEASE_ASSERT(!runScheduleWork.simulationWork.useGpuPme
1789 || (deviceStreamManager != nullptr),
1790 "Device stream manager should be valid in order to use GPU "
1793 !runScheduleWork.simulationWork.useGpuPme
1794 || deviceStreamManager->streamIsValid(DeviceStreamType::Pme),
1795 "GPU PME stream should be valid in order to use GPU version of PME.");
1797 const DeviceContext* deviceContext = runScheduleWork.simulationWork.useGpuPme
1798 ? &deviceStreamManager->context()
1800 const DeviceStream* pmeStream =
1801 runScheduleWork.simulationWork.useGpuPme
1802 ? &deviceStreamManager->stream(DeviceStreamType::Pme)
1805 pmedata = gmx_pme_init(cr,
1806 getNumPmeDomains(cr->dd),
1808 nChargePerturbed != 0,
1809 nTypePerturbed != 0,
1810 mdrunOptions.reproducible,
1813 gmx_omp_nthreads_get(ModuleMultiThread::Pme),
1818 pmeGpuProgram.get(),
1821 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
1826 if (EI_DYNAMICS(inputrec->eI))
1828 /* Turn on signal handling on all nodes */
1830 * (A user signal from the PME nodes (if any)
1831 * is communicated to the PP nodes.
1833 signal_handler_install();
1836 pull_t* pull_work = nullptr;
1837 if (thisRankHasDuty(cr, DUTY_PP))
1839 /* Assumes uniform use of the number of OpenMP threads */
1840 walltime_accounting = walltime_accounting_init(gmx_omp_nthreads_get(ModuleMultiThread::Default));
1842 if (inputrec->bPull)
1844 /* Initialize pull code */
1845 pull_work = init_pull(fplog,
1846 inputrec->pull.get(),
1851 inputrec->fepvals->init_lambda);
1852 if (inputrec->pull->bXOutAverage || inputrec->pull->bFOutAverage)
1854 initPullHistory(pull_work, &observablesHistory);
1856 if (EI_DYNAMICS(inputrec->eI) && MASTER(cr))
1858 init_pull_output_files(pull_work, filenames.size(), filenames.data(), oenv, startingBehavior);
1862 std::unique_ptr<EnforcedRotation> enforcedRotation;
1865 /* Initialize enforced rotation code */
1866 enforcedRotation = init_rot(fplog,
1879 t_swap* swap = nullptr;
1880 if (inputrec->eSwapCoords != SwapType::No)
1882 /* Initialize ion swapping code */
1883 swap = init_swapcoords(fplog,
1885 opt2fn_master("-swap", filenames.size(), filenames.data(), cr),
1888 &observablesHistory,
1896 /* Let makeConstraints know whether we have essential dynamics constraints. */
1897 auto constr = makeConstraints(mtop,
1900 doEssentialDynamics,
1903 updateGroups.useUpdateGroups(),
1909 /* Energy terms and groups */
1910 gmx_enerdata_t enerd(mtop.groups.groups[SimulationAtomGroupType::EnergyOutput].size(),
1911 inputrec->fepvals->n_lambda);
1913 // cos acceleration is only supported by md, but older tpr
1914 // files might still combine it with other integrators
1915 GMX_RELEASE_ASSERT(inputrec->cos_accel == 0.0 || inputrec->eI == IntegrationAlgorithm::MD,
1916 "cos_acceleration is only supported by integrator=md");
1918 /* Kinetic energy data */
1919 gmx_ekindata_t ekind(inputrec->opts.ngtc,
1920 inputrec->cos_accel,
1921 gmx_omp_nthreads_get(ModuleMultiThread::Update));
1923 /* Set up interactive MD (IMD) */
1924 auto imdSession = makeImdSession(inputrec.get(),
1931 MASTER(cr) ? globalState->x : gmx::ArrayRef<gmx::RVec>(),
1935 mdrunOptions.imdOptions,
1938 if (DOMAINDECOMP(cr))
1940 GMX_RELEASE_ASSERT(fr, "fr was NULL while cr->duty was DUTY_PP");
1941 /* This call is not included in init_domain_decomposition
1942 * because fr->atomInfoForEachMoleculeBlock is set later.
1944 makeBondedLinks(cr->dd, mtop, fr->atomInfoForEachMoleculeBlock);
1947 if (runScheduleWork.simulationWork.useGpuBufferOps)
1949 fr->gpuForceReduction[gmx::AtomLocality::Local] = std::make_unique<gmx::GpuForceReduction>(
1950 deviceStreamManager->context(),
1951 deviceStreamManager->stream(gmx::DeviceStreamType::NonBondedLocal),
1953 fr->gpuForceReduction[gmx::AtomLocality::NonLocal] = std::make_unique<gmx::GpuForceReduction>(
1954 deviceStreamManager->context(),
1955 deviceStreamManager->stream(gmx::DeviceStreamType::NonBondedNonLocal),
1959 std::unique_ptr<gmx::StatePropagatorDataGpu> stateGpu;
1960 if (gpusWereDetected
1961 && ((runScheduleWork.simulationWork.useGpuPme && thisRankHasDuty(cr, DUTY_PME))
1962 || runScheduleWork.simulationWork.useGpuBufferOps))
1964 GpuApiCallBehavior transferKind =
1965 (inputrec->eI == IntegrationAlgorithm::MD && !doRerun && !useModularSimulator)
1966 ? GpuApiCallBehavior::Async
1967 : GpuApiCallBehavior::Sync;
1968 GMX_RELEASE_ASSERT(deviceStreamManager != nullptr,
1969 "GPU device stream manager should be initialized to use GPU.");
1970 stateGpu = std::make_unique<gmx::StatePropagatorDataGpu>(
1971 *deviceStreamManager, transferKind, pme_gpu_get_block_size(fr->pmedata), wcycle.get());
1972 fr->stateGpu = stateGpu.get();
1975 GMX_ASSERT(stopHandlerBuilder_, "Runner must provide StopHandlerBuilder to simulator.");
1976 SimulatorBuilder simulatorBuilder;
1978 simulatorBuilder.add(SimulatorStateData(globalState.get(), &observablesHistory, &enerd, &ekind));
1979 simulatorBuilder.add(std::move(membedHolder));
1980 simulatorBuilder.add(std::move(stopHandlerBuilder_));
1981 simulatorBuilder.add(SimulatorConfig(mdrunOptions, startingBehavior, &runScheduleWork));
1984 simulatorBuilder.add(SimulatorEnv(fplog, cr, ms, mdlog, oenv));
1985 simulatorBuilder.add(Profiling(&nrnb, walltime_accounting, wcycle.get()));
1986 simulatorBuilder.add(ConstraintsParam(
1987 constr.get(), enforcedRotation ? enforcedRotation->getLegacyEnfrot() : nullptr, vsite.get()));
1988 // TODO: Separate `fr` to a separate add, and make the `build` handle the coupling sensibly.
1989 simulatorBuilder.add(LegacyInput(
1990 static_cast<int>(filenames.size()), filenames.data(), inputrec.get(), fr.get()));
1991 simulatorBuilder.add(ReplicaExchangeParameters(replExParams));
1992 simulatorBuilder.add(InteractiveMD(imdSession.get()));
1993 simulatorBuilder.add(SimulatorModules(mdModules_->outputProvider(), mdModules_->notifiers()));
1994 simulatorBuilder.add(CenterOfMassPulling(pull_work));
1995 // Todo move to an MDModule
1996 simulatorBuilder.add(IonSwapping(swap));
1997 simulatorBuilder.add(TopologyData(mtop, mdAtoms.get()));
1998 simulatorBuilder.add(BoxDeformationHandle(deform.get()));
1999 simulatorBuilder.add(std::move(modularSimulatorCheckpointData));
2001 // build and run simulator object based on user-input
2002 auto simulator = simulatorBuilder.build(useModularSimulator);
2005 if (fr->pmePpCommGpu)
2007 // destroy object since it is no longer required. (This needs to be done while the GPU context still exists.)
2008 fr->pmePpCommGpu.reset();
2011 if (inputrec->bPull)
2013 finish_pull(pull_work);
2015 finish_swapcoords(swap);
2019 GMX_RELEASE_ASSERT(pmedata, "pmedata was NULL while cr->duty was not DUTY_PP");
2021 walltime_accounting = walltime_accounting_init(gmx_omp_nthreads_get(ModuleMultiThread::Pme));
2022 gmx_pmeonly(pmedata,
2026 walltime_accounting,
2029 runScheduleWork.simulationWork.useGpuPmePpCommunication,
2030 deviceStreamManager.get());
2033 wallcycle_stop(wcycle.get(), WallCycleCounter::Run);
2035 /* Finish up, write some stuff
2036 * if rerunMD, don't write last frame again
2044 walltime_accounting,
2045 fr ? fr->nbv.get() : nullptr,
2047 EI_DYNAMICS(inputrec->eI) && !isMultiSim(ms));
2050 deviceStreamManager.reset(nullptr);
2054 gmx_pme_destroy(pmedata);
2058 // FIXME: this is only here to manually unpin mdAtoms->chargeA_ and state->x,
2059 // before we destroy the GPU context(s)
2060 // Pinned buffers are associated with contexts in CUDA.
2061 // As soon as we destroy GPU contexts after mdrunner() exits, these lines should go.
2062 mdAtoms.reset(nullptr);
2063 globalState.reset(nullptr);
2064 mdModules_.reset(nullptr); // destruct force providers here as they might also use the GPU
2065 listedForcesGpu.reset(nullptr);
2066 fr.reset(nullptr); // destruct forcerec before gpu
2067 // TODO convert to C++ so we can get rid of these frees
2070 if (!hwinfo_->deviceInfoList.empty())
2072 /* stop the GPU profiler (only CUDA) */
2076 /* With tMPI we need to wait for all ranks to finish deallocation before
2077 * destroying the CUDA context as some tMPI ranks may be sharing
2080 * This is not a concern in OpenCL where we use one context per rank.
2082 * Note: it is safe to not call the barrier on the ranks which do not use GPU,
2083 * but it is easier and more futureproof to call it on the whole node.
2085 * Note that this function needs to be called even if GPUs are not used
2086 * in this run because the PME ranks have no knowledge of whether GPUs
2087 * are used or not, but all ranks need to enter the barrier below.
2088 * \todo Remove this physical node barrier after making sure
2089 * that it's not needed anymore (with a shared GPU run).
2093 physicalNodeComm.barrier();
2096 if (!devFlags.usingCudaAwareMpi)
2098 // Don't reset GPU in case of CUDA-AWARE MPI
2099 // UCX creates CUDA buffers which are cleaned-up as part of MPI_Finalize()
2100 // resetting the device before MPI_Finalize() results in crashes inside UCX
2101 releaseDevice(deviceInfo);
2104 /* Does what it says */
2105 print_date_and_time(fplog, cr->nodeid, "Finished mdrun", gmx_gettime());
2106 walltime_accounting_destroy(walltime_accounting);
2108 // Ensure log file content is written
2111 gmx_fio_flush(logFileHandle);
2114 /* Reset FPEs (important for unit tests) by disabling them. Assumes no
2115 * exceptions were enabled before function was called. */
2118 gmx_fedisableexcept();
2121 auto rc = static_cast<int>(gmx_get_stop_condition());
2124 /* we need to join all threads. The sub-threads join when they
2125 exit this function, but the master thread needs to be told to
2135 Mdrunner::~Mdrunner()
2137 // Clean up of the Manager.
2138 // This will end up getting called on every thread-MPI rank, which is unnecessary,
2139 // but okay as long as threads synchronize some time before adding or accessing
2140 // a new set of restraints.
2141 if (restraintManager_)
2143 restraintManager_->clear();
2144 GMX_ASSERT(restraintManager_->countRestraints() == 0,
2145 "restraints added during runner life time should be cleared at runner "
2150 void Mdrunner::addPotential(std::shared_ptr<gmx::IRestraintPotential> puller, const std::string& name)
2152 GMX_ASSERT(restraintManager_, "Mdrunner must have a restraint manager.");
2153 // Not sure if this should be logged through the md logger or something else,
2154 // but it is helpful to have some sort of INFO level message sent somewhere.
2155 // std::cout << "Registering restraint named " << name << std::endl;
2157 // When multiple restraints are used, it may be wasteful to register them separately.
2158 // Maybe instead register an entire Restraint Manager as a force provider.
2159 restraintManager_->addToSpec(std::move(puller), name);
2162 Mdrunner::Mdrunner(std::unique_ptr<MDModules> mdModules) : mdModules_(std::move(mdModules)) {}
2164 Mdrunner::Mdrunner(Mdrunner&&) noexcept = default;
2166 //NOLINTNEXTLINE(performance-noexcept-move-constructor) working around GCC bug 58265 in CentOS 7
2167 Mdrunner& Mdrunner::operator=(Mdrunner&& /*handle*/) noexcept(BUGFREE_NOEXCEPT_STRING) = default;
2169 class Mdrunner::BuilderImplementation
2172 BuilderImplementation() = delete;
2173 BuilderImplementation(std::unique_ptr<MDModules> mdModules, compat::not_null<SimulationContext*> context);
2174 ~BuilderImplementation();
2176 BuilderImplementation& setExtraMdrunOptions(const MdrunOptions& options,
2177 real forceWarningThreshold,
2178 StartingBehavior startingBehavior);
2180 void addHardwareDetectionResult(const gmx_hw_info_t* hwinfo);
2182 void addDomdec(const DomdecOptions& options);
2184 void addInput(SimulationInputHandle inputHolder);
2186 void addVerletList(int nstlist);
2188 void addReplicaExchange(const ReplicaExchangeParameters& params);
2190 void addNonBonded(const char* nbpu_opt);
2192 void addPME(const char* pme_opt_, const char* pme_fft_opt_);
2194 void addBondedTaskAssignment(const char* bonded_opt);
2196 void addUpdateTaskAssignment(const char* update_opt);
2198 void addHardwareOptions(const gmx_hw_opt_t& hardwareOptions);
2200 void addFilenames(ArrayRef<const t_filenm> filenames);
2202 void addOutputEnvironment(gmx_output_env_t* outputEnvironment);
2204 void addLogFile(t_fileio* logFileHandle);
2206 void addStopHandlerBuilder(std::unique_ptr<StopHandlerBuilder> builder);
2211 // Default parameters copied from runner.h
2212 // \todo Clarify source(s) of default parameters.
2214 const char* nbpu_opt_ = nullptr;
2215 const char* pme_opt_ = nullptr;
2216 const char* pme_fft_opt_ = nullptr;
2217 const char* bonded_opt_ = nullptr;
2218 const char* update_opt_ = nullptr;
2220 MdrunOptions mdrunOptions_;
2222 DomdecOptions domdecOptions_;
2224 ReplicaExchangeParameters replicaExchangeParameters_;
2226 //! Command-line override for the duration of a neighbor list with the Verlet scheme.
2229 //! World communicator, used for hardware detection and task assignment
2230 MPI_Comm libraryWorldCommunicator_ = MPI_COMM_NULL;
2232 //! Multisim communicator handle.
2233 gmx_multisim_t* multiSimulation_;
2235 //! mdrun communicator
2236 MPI_Comm simulationCommunicator_ = MPI_COMM_NULL;
2238 //! Print a warning if any force is larger than this (in kJ/mol nm).
2239 real forceWarningThreshold_ = -1;
2241 //! Whether the simulation will start afresh, or restart with/without appending.
2242 StartingBehavior startingBehavior_ = StartingBehavior::NewSimulation;
2244 //! The modules that comprise the functionality of mdrun.
2245 std::unique_ptr<MDModules> mdModules_;
2247 //! Detected hardware.
2248 const gmx_hw_info_t* hwinfo_ = nullptr;
2250 //! \brief Parallelism information.
2251 gmx_hw_opt_t hardwareOptions_;
2253 //! filename options for simulation.
2254 ArrayRef<const t_filenm> filenames_;
2256 /*! \brief Handle to output environment.
2258 * \todo gmx_output_env_t needs lifetime management.
2260 gmx_output_env_t* outputEnvironment_ = nullptr;
2262 /*! \brief Non-owning handle to MD log file.
2264 * \todo Context should own output facilities for client.
2265 * \todo Improve log file handle management.
2267 * Code managing the FILE* relies on the ability to set it to
2268 * nullptr to check whether the filehandle is valid.
2270 t_fileio* logFileHandle_ = nullptr;
2273 * \brief Builder for simulation stop signal handler.
2275 std::unique_ptr<StopHandlerBuilder> stopHandlerBuilder_ = nullptr;
2278 * \brief Sources for initial simulation state.
2280 * See issue #3652 for near-term refinements to the SimulationInput interface.
2282 * See issue #3379 for broader discussion on API aspects of simulation inputs and outputs.
2284 SimulationInputHandle inputHolder_;
2287 Mdrunner::BuilderImplementation::BuilderImplementation(std::unique_ptr<MDModules> mdModules,
2288 compat::not_null<SimulationContext*> context) :
2289 mdModules_(std::move(mdModules))
2291 libraryWorldCommunicator_ = context->libraryWorldCommunicator_;
2292 simulationCommunicator_ = context->simulationCommunicator_;
2293 multiSimulation_ = context->multiSimulation_.get();
2296 Mdrunner::BuilderImplementation::~BuilderImplementation() = default;
2298 Mdrunner::BuilderImplementation&
2299 Mdrunner::BuilderImplementation::setExtraMdrunOptions(const MdrunOptions& options,
2300 const real forceWarningThreshold,
2301 const StartingBehavior startingBehavior)
2303 mdrunOptions_ = options;
2304 forceWarningThreshold_ = forceWarningThreshold;
2305 startingBehavior_ = startingBehavior;
2309 void Mdrunner::BuilderImplementation::addDomdec(const DomdecOptions& options)
2311 domdecOptions_ = options;
2314 void Mdrunner::BuilderImplementation::addVerletList(int nstlist)
2319 void Mdrunner::BuilderImplementation::addReplicaExchange(const ReplicaExchangeParameters& params)
2321 replicaExchangeParameters_ = params;
2324 Mdrunner Mdrunner::BuilderImplementation::build()
2326 auto newRunner = Mdrunner(std::move(mdModules_));
2328 newRunner.mdrunOptions = mdrunOptions_;
2329 newRunner.pforce = forceWarningThreshold_;
2330 newRunner.startingBehavior = startingBehavior_;
2331 newRunner.domdecOptions = domdecOptions_;
2333 // \todo determine an invariant to check or confirm that all gmx_hw_opt_t objects are valid
2334 newRunner.hw_opt = hardwareOptions_;
2336 // No invariant to check. This parameter exists to optionally override other behavior.
2337 newRunner.nstlist_cmdline = nstlist_;
2339 newRunner.replExParams = replicaExchangeParameters_;
2341 newRunner.filenames = filenames_;
2343 newRunner.libraryWorldCommunicator = libraryWorldCommunicator_;
2345 newRunner.simulationCommunicator = simulationCommunicator_;
2347 // nullptr is a valid value for the multisim handle
2348 newRunner.ms = multiSimulation_;
2352 newRunner.hwinfo_ = hwinfo_;
2356 GMX_THROW(gmx::APIError(
2357 "MdrunnerBuilder::addHardwareDetectionResult() is required before build()"));
2362 newRunner.inputHolder_ = std::move(inputHolder_);
2366 GMX_THROW(gmx::APIError("MdrunnerBuilder::addInput() is required before build()."));
2369 // \todo Clarify ownership and lifetime management for gmx_output_env_t
2370 // \todo Update sanity checking when output environment has clearly specified invariants.
2371 // Initialization and default values for oenv are not well specified in the current version.
2372 if (outputEnvironment_)
2374 newRunner.oenv = outputEnvironment_;
2378 GMX_THROW(gmx::APIError(
2379 "MdrunnerBuilder::addOutputEnvironment() is required before build()"));
2382 newRunner.logFileHandle = logFileHandle_;
2386 newRunner.nbpu_opt = nbpu_opt_;
2390 GMX_THROW(gmx::APIError("MdrunnerBuilder::addNonBonded() is required before build()"));
2393 if (pme_opt_ && pme_fft_opt_)
2395 newRunner.pme_opt = pme_opt_;
2396 newRunner.pme_fft_opt = pme_fft_opt_;
2400 GMX_THROW(gmx::APIError("MdrunnerBuilder::addElectrostatics() is required before build()"));
2405 newRunner.bonded_opt = bonded_opt_;
2409 GMX_THROW(gmx::APIError(
2410 "MdrunnerBuilder::addBondedTaskAssignment() is required before build()"));
2415 newRunner.update_opt = update_opt_;
2419 GMX_THROW(gmx::APIError(
2420 "MdrunnerBuilder::addUpdateTaskAssignment() is required before build() "));
2424 newRunner.restraintManager_ = std::make_unique<gmx::RestraintManager>();
2426 if (stopHandlerBuilder_)
2428 newRunner.stopHandlerBuilder_ = std::move(stopHandlerBuilder_);
2432 newRunner.stopHandlerBuilder_ = std::make_unique<StopHandlerBuilder>();
2438 void Mdrunner::BuilderImplementation::addHardwareDetectionResult(const gmx_hw_info_t* hwinfo)
2443 void Mdrunner::BuilderImplementation::addNonBonded(const char* nbpu_opt)
2445 nbpu_opt_ = nbpu_opt;
2448 void Mdrunner::BuilderImplementation::addPME(const char* pme_opt, const char* pme_fft_opt)
2451 pme_fft_opt_ = pme_fft_opt;
2454 void Mdrunner::BuilderImplementation::addBondedTaskAssignment(const char* bonded_opt)
2456 bonded_opt_ = bonded_opt;
2459 void Mdrunner::BuilderImplementation::addUpdateTaskAssignment(const char* update_opt)
2461 update_opt_ = update_opt;
2464 void Mdrunner::BuilderImplementation::addHardwareOptions(const gmx_hw_opt_t& hardwareOptions)
2466 hardwareOptions_ = hardwareOptions;
2469 void Mdrunner::BuilderImplementation::addFilenames(ArrayRef<const t_filenm> filenames)
2471 filenames_ = filenames;
2474 void Mdrunner::BuilderImplementation::addOutputEnvironment(gmx_output_env_t* outputEnvironment)
2476 outputEnvironment_ = outputEnvironment;
2479 void Mdrunner::BuilderImplementation::addLogFile(t_fileio* logFileHandle)
2481 logFileHandle_ = logFileHandle;
2484 void Mdrunner::BuilderImplementation::addStopHandlerBuilder(std::unique_ptr<StopHandlerBuilder> builder)
2486 stopHandlerBuilder_ = std::move(builder);
2489 void Mdrunner::BuilderImplementation::addInput(SimulationInputHandle inputHolder)
2491 inputHolder_ = std::move(inputHolder);
2494 MdrunnerBuilder::MdrunnerBuilder(std::unique_ptr<MDModules> mdModules,
2495 compat::not_null<SimulationContext*> context) :
2496 impl_{ std::make_unique<Mdrunner::BuilderImplementation>(std::move(mdModules), context) }
2500 MdrunnerBuilder::~MdrunnerBuilder() = default;
2502 MdrunnerBuilder& MdrunnerBuilder::addHardwareDetectionResult(const gmx_hw_info_t* hwinfo)
2504 impl_->addHardwareDetectionResult(hwinfo);
2508 MdrunnerBuilder& MdrunnerBuilder::addSimulationMethod(const MdrunOptions& options,
2509 real forceWarningThreshold,
2510 const StartingBehavior startingBehavior)
2512 impl_->setExtraMdrunOptions(options, forceWarningThreshold, startingBehavior);
2516 MdrunnerBuilder& MdrunnerBuilder::addDomainDecomposition(const DomdecOptions& options)
2518 impl_->addDomdec(options);
2522 MdrunnerBuilder& MdrunnerBuilder::addNeighborList(int nstlist)
2524 impl_->addVerletList(nstlist);
2528 MdrunnerBuilder& MdrunnerBuilder::addReplicaExchange(const ReplicaExchangeParameters& params)
2530 impl_->addReplicaExchange(params);
2534 MdrunnerBuilder& MdrunnerBuilder::addNonBonded(const char* nbpu_opt)
2536 impl_->addNonBonded(nbpu_opt);
2540 MdrunnerBuilder& MdrunnerBuilder::addElectrostatics(const char* pme_opt, const char* pme_fft_opt)
2542 // The builder method may become more general in the future, but in this version,
2543 // parameters for PME electrostatics are both required and the only parameters
2545 if (pme_opt && pme_fft_opt)
2547 impl_->addPME(pme_opt, pme_fft_opt);
2552 gmx::InvalidInputError("addElectrostatics() arguments must be non-null pointers."));
2557 MdrunnerBuilder& MdrunnerBuilder::addBondedTaskAssignment(const char* bonded_opt)
2559 impl_->addBondedTaskAssignment(bonded_opt);
2563 MdrunnerBuilder& MdrunnerBuilder::addUpdateTaskAssignment(const char* update_opt)
2565 impl_->addUpdateTaskAssignment(update_opt);
2569 Mdrunner MdrunnerBuilder::build()
2571 return impl_->build();
2574 MdrunnerBuilder& MdrunnerBuilder::addHardwareOptions(const gmx_hw_opt_t& hardwareOptions)
2576 impl_->addHardwareOptions(hardwareOptions);
2580 MdrunnerBuilder& MdrunnerBuilder::addFilenames(ArrayRef<const t_filenm> filenames)
2582 impl_->addFilenames(filenames);
2586 MdrunnerBuilder& MdrunnerBuilder::addOutputEnvironment(gmx_output_env_t* outputEnvironment)
2588 impl_->addOutputEnvironment(outputEnvironment);
2592 MdrunnerBuilder& MdrunnerBuilder::addLogFile(t_fileio* logFileHandle)
2594 impl_->addLogFile(logFileHandle);
2598 MdrunnerBuilder& MdrunnerBuilder::addStopHandlerBuilder(std::unique_ptr<StopHandlerBuilder> builder)
2600 impl_->addStopHandlerBuilder(std::move(builder));
2604 MdrunnerBuilder& MdrunnerBuilder::addInput(SimulationInputHandle input)
2606 impl_->addInput(std::move(input));
2610 MdrunnerBuilder::MdrunnerBuilder(MdrunnerBuilder&&) noexcept = default;
2612 MdrunnerBuilder& MdrunnerBuilder::operator=(MdrunnerBuilder&&) noexcept = default;