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/partition.h"
66 #include "gromacs/ewald/ewald_utils.h"
67 #include "gromacs/ewald/pme.h"
68 #include "gromacs/ewald/pme_gpu_program.h"
69 #include "gromacs/ewald/pme_only.h"
70 #include "gromacs/ewald/pme_pp_comm_gpu.h"
71 #include "gromacs/fileio/checkpoint.h"
72 #include "gromacs/fileio/gmxfio.h"
73 #include "gromacs/fileio/oenv.h"
74 #include "gromacs/fileio/tpxio.h"
75 #include "gromacs/gmxlib/network.h"
76 #include "gromacs/gmxlib/nrnb.h"
77 #include "gromacs/gpu_utils/device_stream_manager.h"
78 #include "gromacs/hardware/cpuinfo.h"
79 #include "gromacs/hardware/detecthardware.h"
80 #include "gromacs/hardware/device_management.h"
81 #include "gromacs/hardware/hardwaretopology.h"
82 #include "gromacs/hardware/printhardware.h"
83 #include "gromacs/imd/imd.h"
84 #include "gromacs/listed_forces/disre.h"
85 #include "gromacs/listed_forces/gpubonded.h"
86 #include "gromacs/listed_forces/listed_forces.h"
87 #include "gromacs/listed_forces/orires.h"
88 #include "gromacs/math/functions.h"
89 #include "gromacs/math/utilities.h"
90 #include "gromacs/math/vec.h"
91 #include "gromacs/mdlib/boxdeformation.h"
92 #include "gromacs/mdlib/broadcaststructs.h"
93 #include "gromacs/mdlib/calc_verletbuf.h"
94 #include "gromacs/mdlib/dispersioncorrection.h"
95 #include "gromacs/mdlib/enerdata_utils.h"
96 #include "gromacs/mdlib/force.h"
97 #include "gromacs/mdlib/forcerec.h"
98 #include "gromacs/mdlib/gmx_omp_nthreads.h"
99 #include "gromacs/mdlib/gpuforcereduction.h"
100 #include "gromacs/mdlib/makeconstraints.h"
101 #include "gromacs/mdlib/md_support.h"
102 #include "gromacs/mdlib/mdatoms.h"
103 #include "gromacs/mdlib/sighandler.h"
104 #include "gromacs/mdlib/stophandler.h"
105 #include "gromacs/mdlib/tgroup.h"
106 #include "gromacs/mdlib/updategroups.h"
107 #include "gromacs/mdlib/vsite.h"
108 #include "gromacs/mdrun/mdmodules.h"
109 #include "gromacs/mdrun/simulationcontext.h"
110 #include "gromacs/mdrun/simulationinput.h"
111 #include "gromacs/mdrun/simulationinputhandle.h"
112 #include "gromacs/mdrunutility/handlerestart.h"
113 #include "gromacs/mdrunutility/logging.h"
114 #include "gromacs/mdrunutility/multisim.h"
115 #include "gromacs/mdrunutility/printtime.h"
116 #include "gromacs/mdrunutility/threadaffinity.h"
117 #include "gromacs/mdtypes/checkpointdata.h"
118 #include "gromacs/mdtypes/commrec.h"
119 #include "gromacs/mdtypes/enerdata.h"
120 #include "gromacs/mdtypes/fcdata.h"
121 #include "gromacs/mdtypes/forcerec.h"
122 #include "gromacs/mdtypes/group.h"
123 #include "gromacs/mdtypes/inputrec.h"
124 #include "gromacs/mdtypes/interaction_const.h"
125 #include "gromacs/mdtypes/md_enums.h"
126 #include "gromacs/mdtypes/mdatom.h"
127 #include "gromacs/mdtypes/mdrunoptions.h"
128 #include "gromacs/mdtypes/observableshistory.h"
129 #include "gromacs/mdtypes/simulation_workload.h"
130 #include "gromacs/mdtypes/state.h"
131 #include "gromacs/mdtypes/state_propagator_data_gpu.h"
132 #include "gromacs/modularsimulator/modularsimulator.h"
133 #include "gromacs/nbnxm/gpu_data_mgmt.h"
134 #include "gromacs/nbnxm/nbnxm.h"
135 #include "gromacs/nbnxm/pairlist_tuning.h"
136 #include "gromacs/pbcutil/pbc.h"
137 #include "gromacs/pulling/output.h"
138 #include "gromacs/pulling/pull.h"
139 #include "gromacs/pulling/pull_rotation.h"
140 #include "gromacs/restraint/manager.h"
141 #include "gromacs/restraint/restraintmdmodule.h"
142 #include "gromacs/restraint/restraintpotential.h"
143 #include "gromacs/swap/swapcoords.h"
144 #include "gromacs/taskassignment/decidegpuusage.h"
145 #include "gromacs/taskassignment/decidesimulationworkload.h"
146 #include "gromacs/taskassignment/resourcedivision.h"
147 #include "gromacs/taskassignment/taskassignment.h"
148 #include "gromacs/taskassignment/usergpuids.h"
149 #include "gromacs/timing/gpu_timing.h"
150 #include "gromacs/timing/wallcycle.h"
151 #include "gromacs/timing/wallcyclereporting.h"
152 #include "gromacs/topology/mtop_util.h"
153 #include "gromacs/trajectory/trajectoryframe.h"
154 #include "gromacs/utility/basenetwork.h"
155 #include "gromacs/utility/cstringutil.h"
156 #include "gromacs/utility/exceptions.h"
157 #include "gromacs/utility/fatalerror.h"
158 #include "gromacs/utility/filestream.h"
159 #include "gromacs/utility/gmxassert.h"
160 #include "gromacs/utility/gmxmpi.h"
161 #include "gromacs/utility/keyvaluetree.h"
162 #include "gromacs/utility/logger.h"
163 #include "gromacs/utility/loggerbuilder.h"
164 #include "gromacs/utility/mdmodulenotification.h"
165 #include "gromacs/utility/physicalnodecommunicator.h"
166 #include "gromacs/utility/pleasecite.h"
167 #include "gromacs/utility/programcontext.h"
168 #include "gromacs/utility/smalloc.h"
169 #include "gromacs/utility/stringutil.h"
170 #include "gromacs/utility/mpiinfo.h"
172 #include "isimulator.h"
173 #include "membedholder.h"
174 #include "replicaexchange.h"
175 #include "simulatorbuilder.h"
181 /*! \brief Manage any development feature flag variables encountered
183 * The use of dev features indicated by environment variables is
184 * logged in order to ensure that runs with such features enabled can
185 * be identified from their log and standard output. Any cross
186 * dependencies are also checked, and if unsatisfied, a fatal error
189 * Note that some development features overrides are applied already here:
190 * the GPU communication flags are set to false in non-tMPI and non-CUDA builds.
192 * \param[in] mdlog Logger object.
193 * \param[in] useGpuForNonbonded True if the nonbonded task is offloaded in this run.
194 * \param[in] pmeRunMode The PME run mode for this run
195 * \returns The object populated with development feature flags.
197 static DevelopmentFeatureFlags manageDevelopmentFeatures(const gmx::MDLogger& mdlog,
198 const bool useGpuForNonbonded,
199 const PmeRunMode pmeRunMode)
201 DevelopmentFeatureFlags devFlags;
203 // Some builds of GCC 5 give false positive warnings that these
204 // getenv results are ignored when clearly they are used.
205 #pragma GCC diagnostic push
206 #pragma GCC diagnostic ignored "-Wunused-result"
208 devFlags.enableGpuBufferOps =
209 GMX_GPU_CUDA && useGpuForNonbonded && (getenv("GMX_USE_GPU_BUFFER_OPS") != nullptr);
210 devFlags.enableGpuHaloExchange = GMX_GPU_CUDA && getenv("GMX_GPU_DD_COMMS") != nullptr;
211 devFlags.forceGpuUpdateDefault = (getenv("GMX_FORCE_UPDATE_DEFAULT_GPU") != nullptr) || GMX_FAHCORE;
212 devFlags.enableGpuPmePPComm =
213 GMX_GPU_CUDA && GMX_THREAD_MPI && getenv("GMX_GPU_PME_PP_COMMS") != nullptr;
215 #pragma GCC diagnostic pop
217 // Direct GPU comm path is being used with CUDA_AWARE_MPI
218 // make sure underlying MPI implementation is CUDA-aware
219 if (!GMX_THREAD_MPI && devFlags.enableGpuHaloExchange)
221 const bool haveDetectedCudaAwareMpi =
222 (checkMpiCudaAwareSupport() == CudaAwareMpiStatus::Supported);
223 const bool forceCudaAwareMpi = (getenv("GMX_FORCE_CUDA_AWARE_MPI") != nullptr);
225 if (!haveDetectedCudaAwareMpi && forceCudaAwareMpi)
227 // CUDA-aware support not detected in MPI library but, user has forced it's use
228 GMX_LOG(mdlog.warning)
230 .appendTextFormatted(
231 "This run has forced use of 'CUDA-aware MPI'. "
232 "But, GROMACS cannot determine if underlying MPI "
233 "is CUDA-aware. GROMACS recommends use of latest openMPI version "
234 "for CUDA-aware support. "
235 "If you observe failures at runtime, try unsetting "
236 "GMX_FORCE_CUDA_AWARE_MPI environment variable.");
239 if (haveDetectedCudaAwareMpi || forceCudaAwareMpi)
241 devFlags.usingCudaAwareMpi = true;
242 GMX_LOG(mdlog.warning)
244 .appendTextFormatted("Using CUDA-aware MPI for 'GPU halo exchange' feature.");
248 if (devFlags.enableGpuHaloExchange)
250 GMX_LOG(mdlog.warning)
252 .appendTextFormatted(
253 "GMX_GPU_DD_COMMS environment variable detected, but the 'GPU "
254 "halo exchange' feature will not be enabled as GROMACS couldn't "
255 "detect CUDA_aware support in underlying MPI implementation.");
256 devFlags.enableGpuHaloExchange = false;
259 GMX_LOG(mdlog.warning)
261 .appendTextFormatted(
262 "GROMACS recommends use of latest OpenMPI version for CUDA-aware "
264 "If you are certain about CUDA-aware support in your MPI library, "
265 "you can force it's use by setting environment variable "
266 " GMX_FORCE_CUDA_AWARE_MPI.");
270 if (devFlags.enableGpuBufferOps)
272 GMX_LOG(mdlog.warning)
274 .appendTextFormatted(
275 "This run uses the 'GPU buffer ops' feature, enabled by the "
276 "GMX_USE_GPU_BUFFER_OPS environment variable.");
279 if (devFlags.forceGpuUpdateDefault)
281 GMX_LOG(mdlog.warning)
283 .appendTextFormatted(
284 "This run will default to '-update gpu' as requested by the "
285 "GMX_FORCE_UPDATE_DEFAULT_GPU environment variable. GPU update with domain "
286 "decomposition lacks substantial testing and should be used with caution.");
289 if (devFlags.enableGpuHaloExchange)
291 if (useGpuForNonbonded)
293 if (!devFlags.enableGpuBufferOps)
295 GMX_LOG(mdlog.warning)
297 .appendTextFormatted(
298 "Enabling GPU buffer operations required by GMX_GPU_DD_COMMS "
299 "(equivalent with GMX_USE_GPU_BUFFER_OPS=1).");
300 devFlags.enableGpuBufferOps = true;
302 GMX_LOG(mdlog.warning)
304 .appendTextFormatted(
305 "This run has requested the 'GPU halo exchange' feature, enabled by "
307 "GMX_GPU_DD_COMMS environment variable.");
311 GMX_LOG(mdlog.warning)
313 .appendTextFormatted(
314 "GMX_GPU_DD_COMMS environment variable detected, but the 'GPU "
315 "halo exchange' feature will not be enabled as nonbonded interactions "
316 "are not offloaded.");
317 devFlags.enableGpuHaloExchange = false;
321 if (devFlags.enableGpuPmePPComm)
323 if (pmeRunMode == PmeRunMode::GPU)
325 if (!devFlags.enableGpuBufferOps)
327 GMX_LOG(mdlog.warning)
329 .appendTextFormatted(
330 "Enabling GPU buffer operations required by GMX_GPU_PME_PP_COMMS "
331 "(equivalent with GMX_USE_GPU_BUFFER_OPS=1).");
332 devFlags.enableGpuBufferOps = true;
334 GMX_LOG(mdlog.warning)
336 .appendTextFormatted(
337 "This run uses the 'GPU PME-PP communications' feature, enabled "
338 "by the GMX_GPU_PME_PP_COMMS environment variable.");
342 std::string clarification;
343 if (pmeRunMode == PmeRunMode::Mixed)
346 "PME FFT and gather are not offloaded to the GPU (PME is running in mixed "
351 clarification = "PME is not offloaded to the GPU.";
353 GMX_LOG(mdlog.warning)
356 "GMX_GPU_PME_PP_COMMS environment variable detected, but the "
357 "'GPU PME-PP communications' feature was not enabled as "
359 devFlags.enableGpuPmePPComm = false;
366 /*! \brief Barrier for safe simultaneous thread access to mdrunner data
368 * Used to ensure that the master thread does not modify mdrunner during copy
369 * on the spawned threads. */
370 static void threadMpiMdrunnerAccessBarrier()
373 MPI_Barrier(MPI_COMM_WORLD);
377 Mdrunner Mdrunner::cloneOnSpawnedThread() const
379 auto newRunner = Mdrunner(std::make_unique<MDModules>());
381 // All runners in the same process share a restraint manager resource because it is
382 // part of the interface to the client code, which is associated only with the
383 // original thread. Handles to the same resources can be obtained by copy.
385 newRunner.restraintManager_ = std::make_unique<RestraintManager>(*restraintManager_);
388 // Copy members of master runner.
389 // \todo Replace with builder when Simulation context and/or runner phases are better defined.
390 // Ref https://gitlab.com/gromacs/gromacs/-/issues/2587 and https://gitlab.com/gromacs/gromacs/-/issues/2375
391 newRunner.hw_opt = hw_opt;
392 newRunner.filenames = filenames;
394 newRunner.hwinfo_ = hwinfo_;
395 newRunner.oenv = oenv;
396 newRunner.mdrunOptions = mdrunOptions;
397 newRunner.domdecOptions = domdecOptions;
398 newRunner.nbpu_opt = nbpu_opt;
399 newRunner.pme_opt = pme_opt;
400 newRunner.pme_fft_opt = pme_fft_opt;
401 newRunner.bonded_opt = bonded_opt;
402 newRunner.update_opt = update_opt;
403 newRunner.nstlist_cmdline = nstlist_cmdline;
404 newRunner.replExParams = replExParams;
405 newRunner.pforce = pforce;
406 // Give the spawned thread the newly created valid communicator
407 // for the simulation.
408 newRunner.libraryWorldCommunicator = MPI_COMM_WORLD;
409 newRunner.simulationCommunicator = MPI_COMM_WORLD;
411 newRunner.startingBehavior = startingBehavior;
412 newRunner.stopHandlerBuilder_ = std::make_unique<StopHandlerBuilder>(*stopHandlerBuilder_);
413 newRunner.inputHolder_ = inputHolder_;
415 threadMpiMdrunnerAccessBarrier();
420 /*! \brief The callback used for running on spawned threads.
422 * Obtains the pointer to the master mdrunner object from the one
423 * argument permitted to the thread-launch API call, copies it to make
424 * a new runner for this thread, reinitializes necessary data, and
425 * proceeds to the simulation. */
426 static void mdrunner_start_fn(const void* arg)
430 auto masterMdrunner = reinterpret_cast<const gmx::Mdrunner*>(arg);
431 /* copy the arg list to make sure that it's thread-local. This
432 doesn't copy pointed-to items, of course; fnm, cr and fplog
433 are reset in the call below, all others should be const. */
434 gmx::Mdrunner mdrunner = masterMdrunner->cloneOnSpawnedThread();
437 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
441 void Mdrunner::spawnThreads(int numThreadsToLaunch)
444 /* now spawn new threads that start mdrunner_start_fn(), while
445 the main thread returns. Thread affinity is handled later. */
446 if (tMPI_Init_fn(TRUE, numThreadsToLaunch, TMPI_AFFINITY_NONE, mdrunner_start_fn, static_cast<const void*>(this))
449 GMX_THROW(gmx::InternalError("Failed to spawn thread-MPI threads"));
452 // Give the master thread the newly created valid communicator for
454 libraryWorldCommunicator = MPI_COMM_WORLD;
455 simulationCommunicator = MPI_COMM_WORLD;
456 threadMpiMdrunnerAccessBarrier();
458 GMX_UNUSED_VALUE(numThreadsToLaunch);
459 GMX_UNUSED_VALUE(mdrunner_start_fn);
465 /*! \brief Initialize variables for Verlet scheme simulation */
466 static void prepare_verlet_scheme(FILE* fplog,
470 const gmx_mtop_t& mtop,
472 bool makeGpuPairList,
473 const gmx::CpuInfo& cpuinfo)
475 // We checked the cut-offs in grompp, but double-check here.
476 // We have PME+LJcutoff kernels for rcoulomb>rvdw.
477 if (EEL_PME_EWALD(ir->coulombtype) && ir->vdwtype == VanDerWaalsType::Cut)
479 GMX_RELEASE_ASSERT(ir->rcoulomb >= ir->rvdw,
480 "With Verlet lists and PME we should have rcoulomb>=rvdw");
484 GMX_RELEASE_ASSERT(ir->rcoulomb == ir->rvdw,
485 "With Verlet lists and no PME rcoulomb and rvdw should be identical");
487 /* For NVE simulations, we will retain the initial list buffer */
488 if (EI_DYNAMICS(ir->eI) && ir->verletbuf_tol > 0
489 && !(EI_MD(ir->eI) && ir->etc == TemperatureCoupling::No))
491 /* Update the Verlet buffer size for the current run setup */
493 /* Here we assume SIMD-enabled kernels are being used. But as currently
494 * calc_verlet_buffer_size gives the same results for 4x8 and 4x4
495 * and 4x2 gives a larger buffer than 4x4, this is ok.
497 ListSetupType listType =
498 (makeGpuPairList ? ListSetupType::Gpu : ListSetupType::CpuSimdWhenSupported);
499 VerletbufListSetup listSetup = verletbufGetSafeListSetup(listType);
501 const real rlist_new =
502 calcVerletBufferSize(mtop, det(box), *ir, ir->nstlist, ir->nstlist - 1, -1, listSetup);
504 if (rlist_new != ir->rlist)
506 if (fplog != nullptr)
509 "\nChanging rlist from %g to %g for non-bonded %dx%d atom kernels\n\n",
512 listSetup.cluster_size_i,
513 listSetup.cluster_size_j);
515 ir->rlist = rlist_new;
519 if (nstlist_cmdline > 0 && (!EI_DYNAMICS(ir->eI) || ir->verletbuf_tol <= 0))
522 "Can not set nstlist without %s",
523 !EI_DYNAMICS(ir->eI) ? "dynamics" : "verlet-buffer-tolerance");
526 if (EI_DYNAMICS(ir->eI))
528 /* Set or try nstlist values */
529 increaseNstlist(fplog, cr, ir, nstlist_cmdline, &mtop, box, makeGpuPairList, cpuinfo);
533 /*! \brief Override the nslist value in inputrec
535 * with value passed on the command line (if any)
537 static void override_nsteps_cmdline(const gmx::MDLogger& mdlog, int64_t nsteps_cmdline, t_inputrec* ir)
541 /* override with anything else than the default -2 */
542 if (nsteps_cmdline > -2)
544 char sbuf_steps[STEPSTRSIZE];
545 char sbuf_msg[STRLEN];
547 ir->nsteps = nsteps_cmdline;
548 if (EI_DYNAMICS(ir->eI) && nsteps_cmdline != -1)
551 "Overriding nsteps with value passed on the command line: %s steps, %.3g ps",
552 gmx_step_str(nsteps_cmdline, sbuf_steps),
553 fabs(nsteps_cmdline * ir->delta_t));
558 "Overriding nsteps with value passed on the command line: %s steps",
559 gmx_step_str(nsteps_cmdline, sbuf_steps));
562 GMX_LOG(mdlog.warning).asParagraph().appendText(sbuf_msg);
564 else if (nsteps_cmdline < -2)
566 gmx_fatal(FARGS, "Invalid nsteps value passed on the command line: %" PRId64, nsteps_cmdline);
568 /* Do nothing if nsteps_cmdline == -2 */
574 /*! \brief Return whether GPU acceleration of nonbondeds is supported with the given settings.
576 * If not, and if a warning may be issued, logs a warning about
577 * falling back to CPU code. With thread-MPI, only the first
578 * call to this function should have \c issueWarning true. */
579 static bool gpuAccelerationOfNonbondedIsUseful(const MDLogger& mdlog, const t_inputrec& ir, bool issueWarning)
581 bool gpuIsUseful = true;
584 if (ir.opts.ngener - ir.nwall > 1)
586 /* The GPU code does not support more than one energy group.
587 * If the user requested GPUs explicitly, a fatal error is given later.
591 "Multiple energy groups is not implemented for GPUs, falling back to the CPU. "
592 "For better performance, run on the GPU without energy groups and then do "
593 "gmx mdrun -rerun option on the trajectory with an energy group .tpr file.";
599 warning = "TPI is not implemented for GPUs.";
602 if (!gpuIsUseful && issueWarning)
604 GMX_LOG(mdlog.warning).asParagraph().appendText(warning);
610 //! Initializes the logger for mdrun.
611 static gmx::LoggerOwner buildLogger(FILE* fplog, const bool isSimulationMasterRank)
613 gmx::LoggerBuilder builder;
614 if (fplog != nullptr)
616 builder.addTargetFile(gmx::MDLogger::LogLevel::Info, fplog);
618 if (isSimulationMasterRank)
620 builder.addTargetStream(gmx::MDLogger::LogLevel::Warning, &gmx::TextOutputFile::standardError());
622 return builder.build();
625 //! Make a TaskTarget from an mdrun argument string.
626 static TaskTarget findTaskTarget(const char* optionString)
628 TaskTarget returnValue = TaskTarget::Auto;
630 if (strncmp(optionString, "auto", 3) == 0)
632 returnValue = TaskTarget::Auto;
634 else if (strncmp(optionString, "cpu", 3) == 0)
636 returnValue = TaskTarget::Cpu;
638 else if (strncmp(optionString, "gpu", 3) == 0)
640 returnValue = TaskTarget::Gpu;
644 GMX_ASSERT(false, "Option string should have been checked for sanity already");
650 //! Finish run, aggregate data to print performance info.
651 static void finish_run(FILE* fplog,
652 const gmx::MDLogger& mdlog,
654 const t_inputrec& inputrec,
656 gmx_wallcycle_t wcycle,
657 gmx_walltime_accounting_t walltime_accounting,
658 nonbonded_verlet_t* nbv,
659 const gmx_pme_t* pme,
663 double nbfs = 0, mflop = 0;
664 double elapsed_time, elapsed_time_over_all_ranks, elapsed_time_over_all_threads,
665 elapsed_time_over_all_threads_over_all_ranks;
666 /* Control whether it is valid to print a report. Only the
667 simulation master may print, but it should not do so if the run
668 terminated e.g. before a scheduled reset step. This is
669 complicated by the fact that PME ranks are unaware of the
670 reason why they were sent a pmerecvqxFINISH. To avoid
671 communication deadlocks, we always do the communication for the
672 report, even if we've decided not to write the report, because
673 how long it takes to finish the run is not important when we've
674 decided not to report on the simulation performance.
676 Further, we only report performance for dynamical integrators,
677 because those are the only ones for which we plan to
678 consider doing any optimizations. */
679 bool printReport = EI_DYNAMICS(inputrec.eI) && SIMMASTER(cr);
681 if (printReport && !walltime_accounting_get_valid_finish(walltime_accounting))
683 GMX_LOG(mdlog.warning)
685 .appendText("Simulation ended prematurely, no performance report will be written.");
690 std::unique_ptr<t_nrnb> nrnbTotalStorage;
693 nrnbTotalStorage = std::make_unique<t_nrnb>();
694 nrnb_tot = nrnbTotalStorage.get();
696 MPI_Allreduce(nrnb->n.data(), nrnb_tot->n.data(), eNRNB, MPI_DOUBLE, MPI_SUM, cr->mpi_comm_mysim);
704 elapsed_time = walltime_accounting_get_time_since_reset(walltime_accounting);
705 elapsed_time_over_all_threads =
706 walltime_accounting_get_time_since_reset_over_all_threads(walltime_accounting);
710 /* reduce elapsed_time over all MPI ranks in the current simulation */
711 MPI_Allreduce(&elapsed_time, &elapsed_time_over_all_ranks, 1, MPI_DOUBLE, MPI_SUM, cr->mpi_comm_mysim);
712 elapsed_time_over_all_ranks /= cr->nnodes;
713 /* Reduce elapsed_time_over_all_threads over all MPI ranks in the
714 * current simulation. */
715 MPI_Allreduce(&elapsed_time_over_all_threads,
716 &elapsed_time_over_all_threads_over_all_ranks,
725 elapsed_time_over_all_ranks = elapsed_time;
726 elapsed_time_over_all_threads_over_all_ranks = elapsed_time_over_all_threads;
731 print_flop(fplog, nrnb_tot, &nbfs, &mflop);
734 if (thisRankHasDuty(cr, DUTY_PP) && DOMAINDECOMP(cr))
736 print_dd_statistics(cr, inputrec, fplog);
739 /* TODO Move the responsibility for any scaling by thread counts
740 * to the code that handled the thread region, so that there's a
741 * mechanism to keep cycle counting working during the transition
742 * to task parallelism. */
743 int nthreads_pp = gmx_omp_nthreads_get(emntNonbonded);
744 int nthreads_pme = gmx_omp_nthreads_get(emntPME);
745 wallcycle_scale_by_num_threads(
746 wcycle, thisRankHasDuty(cr, DUTY_PME) && !thisRankHasDuty(cr, DUTY_PP), nthreads_pp, nthreads_pme);
747 auto cycle_sum(wallcycle_sum(cr, wcycle));
751 auto nbnxn_gpu_timings =
752 (nbv != nullptr && nbv->useGpu()) ? Nbnxm::gpu_get_timings(nbv->gpu_nbv) : nullptr;
753 gmx_wallclock_gpu_pme_t pme_gpu_timings = {};
755 if (pme_gpu_task_enabled(pme))
757 pme_gpu_get_timings(pme, &pme_gpu_timings);
759 wallcycle_print(fplog,
765 elapsed_time_over_all_ranks,
771 if (EI_DYNAMICS(inputrec.eI))
773 delta_t = inputrec.delta_t;
779 elapsed_time_over_all_threads_over_all_ranks,
780 elapsed_time_over_all_ranks,
781 walltime_accounting_get_nsteps_done_since_reset(walltime_accounting),
789 elapsed_time_over_all_threads_over_all_ranks,
790 elapsed_time_over_all_ranks,
791 walltime_accounting_get_nsteps_done_since_reset(walltime_accounting),
799 int Mdrunner::mdrunner()
802 std::unique_ptr<t_forcerec> fr;
803 real ewaldcoeff_q = 0;
804 real ewaldcoeff_lj = 0;
805 int nChargePerturbed = -1, nTypePerturbed = 0;
806 gmx_wallcycle_t wcycle;
807 gmx_walltime_accounting_t walltime_accounting = nullptr;
808 MembedHolder membedHolder(filenames.size(), filenames.data());
810 /* CAUTION: threads may be started later on in this function, so
811 cr doesn't reflect the final parallel state right now */
814 /* TODO: inputrec should tell us whether we use an algorithm, not a file option */
815 const bool doEssentialDynamics = opt2bSet("-ei", filenames.size(), filenames.data());
816 const bool doRerun = mdrunOptions.rerun;
818 // Handle task-assignment related user options.
819 EmulateGpuNonbonded emulateGpuNonbonded =
820 (getenv("GMX_EMULATE_GPU") != nullptr ? EmulateGpuNonbonded::Yes : EmulateGpuNonbonded::No);
822 std::vector<int> userGpuTaskAssignment;
825 userGpuTaskAssignment = parseUserTaskAssignmentString(hw_opt.userGpuTaskAssignment);
827 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
828 auto nonbondedTarget = findTaskTarget(nbpu_opt);
829 auto pmeTarget = findTaskTarget(pme_opt);
830 auto pmeFftTarget = findTaskTarget(pme_fft_opt);
831 auto bondedTarget = findTaskTarget(bonded_opt);
832 auto updateTarget = findTaskTarget(update_opt);
834 FILE* fplog = nullptr;
835 // If we are appending, we don't write log output because we need
836 // to check that the old log file matches what the checkpoint file
837 // expects. Otherwise, we should start to write log output now if
838 // there is a file ready for it.
839 if (logFileHandle != nullptr && startingBehavior != StartingBehavior::RestartWithAppending)
841 fplog = gmx_fio_getfp(logFileHandle);
843 const bool isSimulationMasterRank = findIsSimulationMasterRank(ms, simulationCommunicator);
844 gmx::LoggerOwner logOwner(buildLogger(fplog, isSimulationMasterRank));
845 gmx::MDLogger mdlog(logOwner.logger());
847 gmx_print_detected_hardware(fplog, isSimulationMasterRank && isMasterSim(ms), mdlog, hwinfo_);
849 std::vector<int> availableDevices =
850 makeListOfAvailableDevices(hwinfo_->deviceInfoList, hw_opt.devicesSelectedByUser);
851 const int numAvailableDevices = gmx::ssize(availableDevices);
853 // Print citation requests after all software/hardware printing
854 pleaseCiteGromacs(fplog);
856 // Note: legacy program logic relies on checking whether these pointers are assigned.
857 // Objects may or may not be allocated later.
858 std::unique_ptr<t_inputrec> inputrec;
859 std::unique_ptr<t_state> globalState;
861 auto partialDeserializedTpr = std::make_unique<PartialDeserializedTprFile>();
863 if (isSimulationMasterRank)
865 // Allocate objects to be initialized by later function calls.
866 /* Only the master rank has the global state */
867 globalState = std::make_unique<t_state>();
868 inputrec = std::make_unique<t_inputrec>();
870 /* Read (nearly) all data required for the simulation
871 * and keep the partly serialized tpr contents to send to other ranks later
873 applyGlobalSimulationState(
874 *inputHolder_.get(), partialDeserializedTpr.get(), globalState.get(), inputrec.get(), &mtop);
877 /* Check and update the hardware options for internal consistency */
878 checkAndUpdateHardwareOptions(
879 mdlog, &hw_opt, isSimulationMasterRank, domdecOptions.numPmeRanks, inputrec.get());
881 if (GMX_THREAD_MPI && isSimulationMasterRank)
883 bool useGpuForNonbonded = false;
884 bool useGpuForPme = false;
887 GMX_RELEASE_ASSERT(inputrec != nullptr, "Keep the compiler happy");
889 // If the user specified the number of ranks, then we must
890 // respect that, but in default mode, we need to allow for
891 // the number of GPUs to choose the number of ranks.
892 auto canUseGpuForNonbonded = buildSupportsNonbondedOnGpu(nullptr);
893 useGpuForNonbonded = decideWhetherToUseGpusForNonbondedWithThreadMpi(
895 numAvailableDevices > 0,
896 userGpuTaskAssignment,
898 canUseGpuForNonbonded,
899 gpuAccelerationOfNonbondedIsUseful(mdlog, *inputrec, GMX_THREAD_MPI),
900 hw_opt.nthreads_tmpi);
901 useGpuForPme = decideWhetherToUseGpusForPmeWithThreadMpi(useGpuForNonbonded,
904 userGpuTaskAssignment,
907 hw_opt.nthreads_tmpi,
908 domdecOptions.numPmeRanks);
910 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
912 /* Determine how many thread-MPI ranks to start.
914 * TODO Over-writing the user-supplied value here does
915 * prevent any possible subsequent checks from working
917 hw_opt.nthreads_tmpi = get_nthreads_mpi(hwinfo_,
925 membedHolder.doMembed());
927 // Now start the threads for thread MPI.
928 spawnThreads(hw_opt.nthreads_tmpi);
929 // The spawned threads enter mdrunner() and execution of
930 // master and spawned threads joins at the end of this block.
933 GMX_RELEASE_ASSERT(ms || simulationCommunicator != MPI_COMM_NULL,
934 "Must have valid communicator unless running a multi-simulation");
935 CommrecHandle crHandle = init_commrec(simulationCommunicator);
936 t_commrec* cr = crHandle.get();
937 GMX_RELEASE_ASSERT(cr != nullptr, "Must have valid commrec");
939 PhysicalNodeCommunicator physicalNodeComm(libraryWorldCommunicator, gmx_physicalnode_id_hash());
941 // If we detected the topology on this system, double-check that it makes sense
942 if (hwinfo_->hardwareTopology->isThisSystem())
944 hardwareTopologyDoubleCheckDetection(mdlog, *hwinfo_->hardwareTopology);
949 /* now broadcast everything to the non-master nodes/threads: */
950 if (!isSimulationMasterRank)
952 // Until now, only the master rank has a non-null pointer.
953 // On non-master ranks, allocate the object that will receive data in the following call.
954 inputrec = std::make_unique<t_inputrec>();
956 init_parallel(cr->mpiDefaultCommunicator,
960 partialDeserializedTpr.get());
962 GMX_RELEASE_ASSERT(inputrec != nullptr, "All ranks should have a valid inputrec now");
963 partialDeserializedTpr.reset(nullptr);
966 !inputrec->useConstantAcceleration,
967 "Linear acceleration has been removed in GROMACS 2022, and was broken for many years "
968 "before that. Use GROMACS 4.5 or earlier if you need this feature.");
970 // Now the number of ranks is known to all ranks, and each knows
971 // the inputrec read by the master rank. The ranks can now all run
972 // the task-deciding functions and will agree on the result
973 // without needing to communicate.
974 const bool useDomainDecomposition =
975 (PAR(cr) && !(EI_TPI(inputrec->eI) || inputrec->eI == IntegrationAlgorithm::NM));
977 // Note that these variables describe only their own node.
979 // Note that when bonded interactions run on a GPU they always run
980 // alongside a nonbonded task, so do not influence task assignment
981 // even though they affect the force calculation workload.
982 bool useGpuForNonbonded = false;
983 bool useGpuForPme = false;
984 bool useGpuForBonded = false;
985 bool useGpuForUpdate = false;
986 bool gpusWereDetected = hwinfo_->ngpu_compatible_tot > 0;
989 // It's possible that there are different numbers of GPUs on
990 // different nodes, which is the user's responsibility to
991 // handle. If unsuitable, we will notice that during task
993 auto canUseGpuForNonbonded = buildSupportsNonbondedOnGpu(nullptr);
994 useGpuForNonbonded = decideWhetherToUseGpusForNonbonded(
996 userGpuTaskAssignment,
998 canUseGpuForNonbonded,
999 gpuAccelerationOfNonbondedIsUseful(mdlog, *inputrec, !GMX_THREAD_MPI),
1001 useGpuForPme = decideWhetherToUseGpusForPme(useGpuForNonbonded,
1003 userGpuTaskAssignment,
1006 cr->sizeOfDefaultCommunicator,
1007 domdecOptions.numPmeRanks,
1009 useGpuForBonded = decideWhetherToUseGpusForBonded(
1010 useGpuForNonbonded, useGpuForPme, bondedTarget, *inputrec, mtop, domdecOptions.numPmeRanks, gpusWereDetected);
1012 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
1014 const PmeRunMode pmeRunMode = determinePmeRunMode(useGpuForPme, pmeFftTarget, *inputrec);
1016 // Initialize development feature flags that enabled by environment variable
1017 // and report those features that are enabled.
1018 const DevelopmentFeatureFlags devFlags =
1019 manageDevelopmentFeatures(mdlog, useGpuForNonbonded, pmeRunMode);
1021 const bool useModularSimulator = checkUseModularSimulator(false,
1028 doEssentialDynamics,
1029 membedHolder.doMembed());
1031 // Build restraints.
1032 // TODO: hide restraint implementation details from Mdrunner.
1033 // There is nothing unique about restraints at this point as far as the
1034 // Mdrunner is concerned. The Mdrunner should just be getting a sequence of
1035 // factory functions from the SimulationContext on which to call mdModules_->add().
1036 // TODO: capture all restraints into a single RestraintModule, passed to the runner builder.
1037 for (auto&& restraint : restraintManager_->getRestraints())
1039 auto module = RestraintMDModule::create(restraint, restraint->sites());
1040 mdModules_->add(std::move(module));
1043 // TODO: Error handling
1044 mdModules_->assignOptionsToModules(*inputrec->params, nullptr);
1045 // now that the MDModules know their options, they know which callbacks to sign up to
1046 mdModules_->subscribeToSimulationSetupNotifications();
1047 const auto& setupNotifier = mdModules_->notifiers().simulationSetupNotifier_;
1049 if (inputrec->internalParameters != nullptr)
1051 setupNotifier.notify(*inputrec->internalParameters);
1054 if (fplog != nullptr)
1056 pr_inputrec(fplog, 0, "Input Parameters", inputrec.get(), FALSE);
1057 fprintf(fplog, "\n");
1062 /* In rerun, set velocities to zero if present */
1063 if (doRerun && ((globalState->flags & enumValueToBitMask(StateEntry::V)) != 0))
1065 // rerun does not use velocities
1069 "Rerun trajectory contains velocities. Rerun does only evaluate "
1070 "potential energy and forces. The velocities will be ignored.");
1071 for (int i = 0; i < globalState->natoms; i++)
1073 clear_rvec(globalState->v[i]);
1075 globalState->flags &= ~enumValueToBitMask(StateEntry::V);
1078 /* now make sure the state is initialized and propagated */
1079 set_state_entries(globalState.get(), inputrec.get(), useModularSimulator);
1082 /* NM and TPI parallelize over force/energy calculations, not atoms,
1083 * so we need to initialize and broadcast the global state.
1085 if (inputrec->eI == IntegrationAlgorithm::NM || inputrec->eI == IntegrationAlgorithm::TPI)
1089 globalState = std::make_unique<t_state>();
1091 broadcastStateWithoutDynamics(
1092 cr->mpiDefaultCommunicator, DOMAINDECOMP(cr), PAR(cr), globalState.get());
1095 /* A parallel command line option consistency check that we can
1096 only do after any threads have started. */
1098 && (domdecOptions.numCells[XX] > 1 || domdecOptions.numCells[YY] > 1
1099 || domdecOptions.numCells[ZZ] > 1 || domdecOptions.numPmeRanks > 0))
1102 "The -dd or -npme option request a parallel simulation, "
1104 "but %s was compiled without threads or MPI enabled",
1105 output_env_get_program_display_name(oenv));
1106 #elif GMX_THREAD_MPI
1107 "but the number of MPI-threads (option -ntmpi) is not set or is 1");
1109 "but %s was not started through mpirun/mpiexec or only one rank was requested "
1110 "through mpirun/mpiexec",
1111 output_env_get_program_display_name(oenv));
1115 if (doRerun && (EI_ENERGY_MINIMIZATION(inputrec->eI) || IntegrationAlgorithm::NM == inputrec->eI))
1118 "The .mdp file specified an energy mininization or normal mode algorithm, and "
1119 "these are not compatible with mdrun -rerun");
1122 /* Object for collecting reasons for not using PME-only ranks */
1123 SeparatePmeRanksPermitted separatePmeRanksPermitted;
1125 /* Permit MDModules to notify whether they want to use PME-only ranks */
1126 setupNotifier.notify(&separatePmeRanksPermitted);
1128 /* If simulation is not using PME then disable PME-only ranks */
1129 if (!(EEL_PME(inputrec->coulombtype) || EVDW_PME(inputrec->vdwtype)))
1131 separatePmeRanksPermitted.disablePmeRanks(
1132 "PME-only ranks are requested, but the system does not use PME "
1133 "for electrostatics or LJ");
1136 /* With NB GPUs we don't automatically use PME-only CPU ranks. PME ranks can
1137 * improve performance with many threads per GPU, since our OpenMP
1138 * scaling is bad, but it's difficult to automate the setup.
1140 if (useGpuForNonbonded && domdecOptions.numPmeRanks < 0)
1142 separatePmeRanksPermitted.disablePmeRanks(
1143 "PME-only CPU ranks are not automatically used when "
1144 "non-bonded interactions are computed on GPUs");
1147 /* If GPU is used for PME then only 1 PME rank is permitted */
1148 if (useGpuForPme && (domdecOptions.numPmeRanks < 0 || domdecOptions.numPmeRanks > 1))
1150 separatePmeRanksPermitted.disablePmeRanks(
1151 "PME GPU decomposition is not supported. Only one separate PME-only GPU rank "
1155 /* Disable PME-only ranks if some parts of the code requested so and it's up to GROMACS to decide */
1156 if (!separatePmeRanksPermitted.permitSeparatePmeRanks() && domdecOptions.numPmeRanks < 0)
1158 domdecOptions.numPmeRanks = 0;
1161 .appendText("Simulation will not use PME-only ranks because: "
1162 + separatePmeRanksPermitted.reasonsWhyDisabled());
1165 /* If some parts of the code could not use PME-only ranks and
1166 * user explicitly used mdrun -npme option then throw an error */
1167 if (!separatePmeRanksPermitted.permitSeparatePmeRanks() && domdecOptions.numPmeRanks > 0)
1169 gmx_fatal_collective(FARGS,
1170 cr->mpiDefaultCommunicator,
1172 "Requested -npme %d option is not viable because: %s",
1173 domdecOptions.numPmeRanks,
1174 separatePmeRanksPermitted.reasonsWhyDisabled().c_str());
1177 /* NMR restraints must be initialized before load_checkpoint,
1178 * since with time averaging the history is added to t_state.
1179 * For proper consistency check we therefore need to extend
1181 * So the PME-only nodes (if present) will also initialize
1182 * the distance restraints.
1185 /* This needs to be called before read_checkpoint to extend the state */
1186 t_disresdata* disresdata;
1187 snew(disresdata, 1);
1191 DisResRunMode::MDRun,
1192 MASTER(cr) ? DDRole::Master : DDRole::Agent,
1193 PAR(cr) ? NumRanks::Multiple : NumRanks::Single,
1198 replExParams.exchangeInterval > 0);
1200 t_oriresdata* oriresdata;
1201 snew(oriresdata, 1);
1202 init_orires(fplog, mtop, inputrec.get(), cr, ms, globalState.get(), oriresdata);
1204 auto deform = prepareBoxDeformation(globalState != nullptr ? globalState->box : box,
1205 MASTER(cr) ? DDRole::Master : DDRole::Agent,
1206 PAR(cr) ? NumRanks::Multiple : NumRanks::Single,
1207 cr->mpi_comm_mygroup,
1211 /* We have to remember the generation's first step before reading checkpoint.
1212 This way, we can report to the F@H core both the generation's first step
1213 and the restored first step, thus making it able to distinguish between
1214 an interruption/resume and start of the n-th generation simulation.
1215 Having this information, the F@H core can correctly calculate and report
1218 int gen_first_step = 0;
1221 gen_first_step = inputrec->init_step;
1225 ObservablesHistory observablesHistory = {};
1227 auto modularSimulatorCheckpointData = std::make_unique<ReadCheckpointDataHolder>();
1228 if (startingBehavior != StartingBehavior::NewSimulation)
1230 /* Check if checkpoint file exists before doing continuation.
1231 * This way we can use identical input options for the first and subsequent runs...
1233 if (mdrunOptions.numStepsCommandline > -2)
1235 /* Temporarily set the number of steps to unlimited to avoid
1236 * triggering the nsteps check in load_checkpoint().
1237 * This hack will go away soon when the -nsteps option is removed.
1239 inputrec->nsteps = -1;
1242 // Finish applying initial simulation state information from external sources on all ranks.
1243 // Reconcile checkpoint file data with Mdrunner state established up to this point.
1244 applyLocalState(*inputHolder_.get(),
1247 domdecOptions.numCells,
1250 &observablesHistory,
1251 mdrunOptions.reproducible,
1252 mdModules_->notifiers(),
1253 modularSimulatorCheckpointData.get(),
1254 useModularSimulator);
1255 // TODO: (#3652) Synchronize filesystem state, SimulationInput contents, and program
1257 // on all code paths.
1258 // Write checkpoint or provide hook to update SimulationInput.
1259 // If there was a checkpoint file, SimulationInput contains more information
1260 // than if there wasn't. At this point, we have synchronized the in-memory
1261 // state with the filesystem state only for restarted simulations. We should
1262 // be calling applyLocalState unconditionally and expect that the completeness
1263 // of SimulationInput is not dependent on its creation method.
1265 if (startingBehavior == StartingBehavior::RestartWithAppending && logFileHandle)
1267 // Now we can start normal logging to the truncated log file.
1268 fplog = gmx_fio_getfp(logFileHandle);
1269 prepareLogAppending(fplog);
1270 logOwner = buildLogger(fplog, MASTER(cr));
1271 mdlog = logOwner.logger();
1278 fcRegisterSteps(inputrec->nsteps + inputrec->init_step, gen_first_step);
1282 if (mdrunOptions.numStepsCommandline > -2)
1287 "The -nsteps functionality is deprecated, and may be removed in a future "
1289 "Consider using gmx convert-tpr -nsteps or changing the appropriate .mdp "
1292 /* override nsteps with value set on the commandline */
1293 override_nsteps_cmdline(mdlog, mdrunOptions.numStepsCommandline, inputrec.get());
1295 if (isSimulationMasterRank)
1297 copy_mat(globalState->box, box);
1302 gmx_bcast(sizeof(box), box, cr->mpiDefaultCommunicator);
1305 if (inputrec->cutoff_scheme != CutoffScheme::Verlet)
1308 "This group-scheme .tpr file can no longer be run by mdrun. Please update to the "
1309 "Verlet scheme, or use an earlier version of GROMACS if necessary.");
1311 /* Update rlist and nstlist. */
1312 /* Note: prepare_verlet_scheme is calling increaseNstlist(...), which (while attempting to
1313 * increase rlist) tries to check if the newly chosen value fits with the DD scheme. As this is
1314 * run before any DD scheme is set up, this check is never executed. See #3334 for more details.
1316 prepare_verlet_scheme(fplog,
1322 useGpuForNonbonded || (emulateGpuNonbonded == EmulateGpuNonbonded::Yes),
1325 // This builder is necessary while we have multi-part construction
1326 // of DD. Before DD is constructed, we use the existence of
1327 // the builder object to indicate that further construction of DD
1329 std::unique_ptr<DomainDecompositionBuilder> ddBuilder;
1330 if (useDomainDecomposition)
1332 ddBuilder = std::make_unique<DomainDecompositionBuilder>(
1340 positionsFromStatePointer(globalState.get()));
1344 /* PME, if used, is done on all nodes with 1D decomposition */
1345 cr->nnodes = cr->sizeOfDefaultCommunicator;
1346 cr->sim_nodeid = cr->rankInDefaultCommunicator;
1347 cr->nodeid = cr->rankInDefaultCommunicator;
1349 cr->duty = (DUTY_PP | DUTY_PME);
1351 if (inputrec->pbcType == PbcType::Screw)
1353 gmx_fatal(FARGS, "pbc=screw is only implemented with domain decomposition");
1357 // Produce the task assignment for this rank - done after DD is constructed
1358 GpuTaskAssignments gpuTaskAssignments = GpuTaskAssignmentsBuilder::build(
1360 userGpuTaskAssignment,
1362 simulationCommunicator,
1370 thisRankHasDuty(cr, DUTY_PP),
1371 // TODO cr->duty & DUTY_PME should imply that a PME
1372 // algorithm is active, but currently does not.
1373 EEL_PME(inputrec->coulombtype) && thisRankHasDuty(cr, DUTY_PME));
1375 // Get the device handles for the modules, nullptr when no task is assigned.
1377 DeviceInformation* deviceInfo = gpuTaskAssignments.initDevice(&deviceId);
1379 // timing enabling - TODO put this in gpu_utils (even though generally this is just option handling?)
1380 bool useTiming = true;
1384 /* WARNING: CUDA timings are incorrect with multiple streams.
1385 * This is the main reason why they are disabled by default.
1387 // TODO: Consider turning on by default when we can detect nr of streams.
1388 useTiming = (getenv("GMX_ENABLE_GPU_TIMING") != nullptr);
1390 else if (GMX_GPU_OPENCL)
1392 useTiming = (getenv("GMX_DISABLE_GPU_TIMING") == nullptr);
1395 // TODO Currently this is always built, yet DD partition code
1396 // checks if it is built before using it. Probably it should
1397 // become an MDModule that is made only when another module
1398 // requires it (e.g. pull, CompEl, density fitting), so that we
1399 // don't update the local atom sets unilaterally every step.
1400 LocalAtomSetManager atomSets;
1403 // TODO Pass the GPU streams to ddBuilder to use in buffer
1404 // transfers (e.g. halo exchange)
1405 cr->dd = ddBuilder->build(&atomSets);
1406 // The builder's job is done, so destruct it
1407 ddBuilder.reset(nullptr);
1408 // Note that local state still does not exist yet.
1411 // The GPU update is decided here because we need to know whether the constraints or
1412 // SETTLEs can span accross the domain borders (i.e. whether or not update groups are
1413 // defined). This is only known after DD is initialized, hence decision on using GPU
1414 // update is done so late.
1417 const bool useUpdateGroups = cr->dd ? ddUsesUpdateGroups(*cr->dd) : false;
1418 const bool haveFrozenAtoms = inputrecFrozenAtoms(inputrec.get());
1420 useGpuForUpdate = decideWhetherToUseGpuForUpdate(useDomainDecomposition,
1423 domdecOptions.numPmeRanks > 0,
1429 doEssentialDynamics,
1430 gmx_mtop_ftype_count(mtop, F_ORIRES) > 0,
1431 replExParams.exchangeInterval > 0,
1437 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
1439 const bool printHostName = (cr->nnodes > 1);
1440 gpuTaskAssignments.reportGpuUsage(mdlog, printHostName, useGpuForBonded, pmeRunMode, useGpuForUpdate);
1442 const bool disableNonbondedCalculation = (getenv("GMX_NO_NONBONDED") != nullptr);
1443 if (disableNonbondedCalculation)
1445 /* turn off non-bonded calculations */
1446 GMX_LOG(mdlog.warning)
1449 "Found environment variable GMX_NO_NONBONDED.\n"
1450 "Disabling nonbonded calculations.");
1453 MdrunScheduleWorkload runScheduleWork;
1455 bool useGpuDirectHalo = decideWhetherToUseGpuForHalo(devFlags,
1456 havePPDomainDecomposition(cr),
1458 useModularSimulator,
1460 EI_ENERGY_MINIMIZATION(inputrec->eI));
1462 // Also populates the simulation constant workload description.
1463 runScheduleWork.simulationWork = createSimulationWorkload(*inputrec,
1464 disableNonbondedCalculation,
1472 std::unique_ptr<DeviceStreamManager> deviceStreamManager = nullptr;
1474 if (deviceInfo != nullptr)
1476 if (DOMAINDECOMP(cr) && thisRankHasDuty(cr, DUTY_PP))
1478 dd_setup_dlb_resource_sharing(cr, deviceId);
1480 deviceStreamManager = std::make_unique<DeviceStreamManager>(
1481 *deviceInfo, havePPDomainDecomposition(cr), runScheduleWork.simulationWork, useTiming);
1484 // If the user chose a task assignment, give them some hints
1485 // where appropriate.
1486 if (!userGpuTaskAssignment.empty())
1488 gpuTaskAssignments.logPerformanceHints(mdlog, numAvailableDevices);
1493 /* After possible communicator splitting in make_dd_communicators.
1494 * we can set up the intra/inter node communication.
1496 gmx_setup_nodecomm(fplog, cr);
1502 GMX_LOG(mdlog.warning)
1504 .appendTextFormatted(
1505 "This is simulation %d out of %d running as a composite GROMACS\n"
1506 "multi-simulation job. Setup for this simulation:\n",
1507 ms->simulationIndex_,
1508 ms->numSimulations_);
1510 GMX_LOG(mdlog.warning)
1511 .appendTextFormatted("Using %d MPI %s\n",
1514 cr->nnodes == 1 ? "thread" : "threads"
1516 cr->nnodes == 1 ? "process" : "processes"
1522 // If mdrun -pin auto honors any affinity setting that already
1523 // exists. If so, it is nice to provide feedback about whether
1524 // that existing affinity setting was from OpenMP or something
1525 // else, so we run this code both before and after we initialize
1526 // the OpenMP support.
1527 gmx_check_thread_affinity_set(mdlog, &hw_opt, hwinfo_->nthreads_hw_avail, FALSE);
1528 /* Check and update the number of OpenMP threads requested */
1529 checkAndUpdateRequestedNumOpenmpThreads(
1530 &hw_opt, *hwinfo_, cr, ms, physicalNodeComm.size_, pmeRunMode, mtop, *inputrec);
1532 gmx_omp_nthreads_init(mdlog,
1534 hwinfo_->nthreads_hw_avail,
1535 physicalNodeComm.size_,
1536 hw_opt.nthreads_omp,
1537 hw_opt.nthreads_omp_pme,
1538 !thisRankHasDuty(cr, DUTY_PP));
1540 const bool bEnableFPE = gmxShouldEnableFPExceptions();
1541 // FIXME - reconcile with gmx_feenableexcept() call from CommandLineModuleManager::run()
1544 gmx_feenableexcept();
1547 /* Now that we know the setup is consistent, check for efficiency */
1548 check_resource_division_efficiency(
1549 hwinfo_, gpuTaskAssignments.thisRankHasAnyGpuTask(), mdrunOptions.ntompOptionIsSet, cr, mdlog);
1551 /* getting number of PP/PME threads on this MPI / tMPI rank.
1552 PME: env variable should be read only on one node to make sure it is
1553 identical everywhere;
1555 const int numThreadsOnThisRank = thisRankHasDuty(cr, DUTY_PP) ? gmx_omp_nthreads_get(emntNonbonded)
1556 : gmx_omp_nthreads_get(emntPME);
1557 checkHardwareOversubscription(
1558 numThreadsOnThisRank, cr->nodeid, *hwinfo_->hardwareTopology, physicalNodeComm, mdlog);
1560 // Enable Peer access between GPUs where available
1561 // Only for DD, only master PP rank needs to perform setup, and only if thread MPI plus
1562 // any of the GPU communication features are active.
1563 if (DOMAINDECOMP(cr) && MASTER(cr) && thisRankHasDuty(cr, DUTY_PP) && GMX_THREAD_MPI
1564 && (runScheduleWork.simulationWork.useGpuHaloExchange
1565 || runScheduleWork.simulationWork.useGpuPmePpCommunication))
1567 setupGpuDevicePeerAccess(gpuTaskAssignments.deviceIdsAssigned(), mdlog);
1570 if (hw_opt.threadAffinity != ThreadAffinity::Off)
1572 /* Before setting affinity, check whether the affinity has changed
1573 * - which indicates that probably the OpenMP library has changed it
1574 * since we first checked).
1576 gmx_check_thread_affinity_set(mdlog, &hw_opt, hwinfo_->nthreads_hw_avail, TRUE);
1578 int numThreadsOnThisNode, intraNodeThreadOffset;
1579 analyzeThreadsOnThisNode(
1580 physicalNodeComm, numThreadsOnThisRank, &numThreadsOnThisNode, &intraNodeThreadOffset);
1582 /* Set the CPU affinity */
1583 gmx_set_thread_affinity(mdlog,
1586 *hwinfo_->hardwareTopology,
1587 numThreadsOnThisRank,
1588 numThreadsOnThisNode,
1589 intraNodeThreadOffset,
1593 if (mdrunOptions.timingOptions.resetStep > -1)
1598 "The -resetstep functionality is deprecated, and may be removed in a "
1601 wcycle = wallcycle_init(fplog, mdrunOptions.timingOptions.resetStep, cr);
1605 /* Master synchronizes its value of reset_counters with all nodes
1606 * including PME only nodes */
1607 int64_t reset_counters = wcycle_get_reset_counters(wcycle);
1608 gmx_bcast(sizeof(reset_counters), &reset_counters, cr->mpi_comm_mysim);
1609 wcycle_set_reset_counters(wcycle, reset_counters);
1612 // Membrane embedding must be initialized before we call init_forcerec()
1613 membedHolder.initializeMembed(fplog,
1620 &mdrunOptions.checkpointOptions.period);
1622 const bool thisRankHasPmeGpuTask = gpuTaskAssignments.thisRankHasPmeGpuTask();
1623 std::unique_ptr<MDAtoms> mdAtoms;
1624 std::unique_ptr<VirtualSitesHandler> vsite;
1625 std::unique_ptr<GpuBonded> gpuBonded;
1628 if (thisRankHasDuty(cr, DUTY_PP))
1630 setupNotifier.notify(*cr);
1631 setupNotifier.notify(&atomSets);
1632 setupNotifier.notify(mtop);
1633 setupNotifier.notify(inputrec->pbcType);
1634 setupNotifier.notify(SimulationTimeStep{ inputrec->delta_t });
1635 /* Initiate forcerecord */
1636 fr = std::make_unique<t_forcerec>();
1637 fr->forceProviders = mdModules_->initForceProviders();
1638 init_forcerec(fplog,
1645 opt2fn("-table", filenames.size(), filenames.data()),
1646 opt2fn("-tablep", filenames.size(), filenames.data()),
1647 opt2fns("-tableb", filenames.size(), filenames.data()),
1649 // Dirty hack, for fixing disres and orires should be made mdmodules
1650 fr->fcdata->disres = disresdata;
1651 fr->fcdata->orires = oriresdata;
1653 // Save a handle to device stream manager to use elsewhere in the code
1654 // TODO: Forcerec is not a correct place to store it.
1655 fr->deviceStreamManager = deviceStreamManager.get();
1657 if (runScheduleWork.simulationWork.useGpuPmePpCommunication && !thisRankHasDuty(cr, DUTY_PME))
1660 deviceStreamManager != nullptr,
1661 "GPU device stream manager should be valid in order to use PME-PP direct "
1664 deviceStreamManager->streamIsValid(DeviceStreamType::PmePpTransfer),
1665 "GPU PP-PME stream should be valid in order to use GPU PME-PP direct "
1667 fr->pmePpCommGpu = std::make_unique<gmx::PmePpCommGpu>(
1670 deviceStreamManager->context(),
1671 deviceStreamManager->stream(DeviceStreamType::PmePpTransfer));
1674 fr->nbv = Nbnxm::init_nb_verlet(mdlog,
1679 runScheduleWork.simulationWork.useGpuNonbonded,
1680 deviceStreamManager.get(),
1684 // TODO: Move the logic below to a GPU bonded builder
1685 if (runScheduleWork.simulationWork.useGpuBonded)
1687 GMX_RELEASE_ASSERT(deviceStreamManager != nullptr,
1688 "GPU device stream manager should be valid in order to use GPU "
1689 "version of bonded forces.");
1690 gpuBonded = std::make_unique<GpuBonded>(
1692 fr->ic->epsfac * fr->fudgeQQ,
1693 deviceStreamManager->context(),
1694 deviceStreamManager->bondedStream(havePPDomainDecomposition(cr)),
1696 fr->gpuBonded = gpuBonded.get();
1699 /* Initialize the mdAtoms structure.
1700 * mdAtoms is not filled with atom data,
1701 * as this can not be done now with domain decomposition.
1703 mdAtoms = makeMDAtoms(fplog, mtop, *inputrec, thisRankHasPmeGpuTask);
1704 if (globalState && thisRankHasPmeGpuTask)
1706 // The pinning of coordinates in the global state object works, because we only use
1707 // PME on GPU without DD or on a separate PME rank, and because the local state pointer
1708 // points to the global state object without DD.
1709 // FIXME: MD and EM separately set up the local state - this should happen in the same
1710 // function, which should also perform the pinning.
1711 changePinningPolicy(&globalState->x, pme_get_pinning_policy());
1714 /* Initialize the virtual site communication */
1715 vsite = makeVirtualSitesHandler(mtop, cr, fr->pbcType);
1717 calc_shifts(box, fr->shift_vec);
1719 /* With periodic molecules the charge groups should be whole at start up
1720 * and the virtual sites should not be far from their proper positions.
1722 if (!inputrec->bContinuation && MASTER(cr)
1723 && !(inputrec->pbcType != PbcType::No && inputrec->bPeriodicMols))
1725 /* Make molecules whole at start of run */
1726 if (fr->pbcType != PbcType::No)
1728 do_pbc_first_mtop(fplog, inputrec->pbcType, box, &mtop, globalState->x.rvec_array());
1732 /* Correct initial vsite positions are required
1733 * for the initial distribution in the domain decomposition
1734 * and for the initial shell prediction.
1736 constructVirtualSitesGlobal(mtop, globalState->x);
1740 if (EEL_PME(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype))
1742 ewaldcoeff_q = fr->ic->ewaldcoeff_q;
1743 ewaldcoeff_lj = fr->ic->ewaldcoeff_lj;
1748 /* This is a PME only node */
1750 GMX_ASSERT(globalState == nullptr,
1751 "We don't need the state on a PME only rank and expect it to be unitialized");
1753 ewaldcoeff_q = calc_ewaldcoeff_q(inputrec->rcoulomb, inputrec->ewald_rtol);
1754 ewaldcoeff_lj = calc_ewaldcoeff_lj(inputrec->rvdw, inputrec->ewald_rtol_lj);
1757 gmx_pme_t* sepPmeData = nullptr;
1758 // This reference hides the fact that PME data is owned by runner on PME-only ranks and by forcerec on other ranks
1759 GMX_ASSERT(thisRankHasDuty(cr, DUTY_PP) == (fr != nullptr),
1760 "Double-checking that only PME-only ranks have no forcerec");
1761 gmx_pme_t*& pmedata = fr ? fr->pmedata : sepPmeData;
1763 // TODO should live in ewald module once its testing is improved
1765 // Later, this program could contain kernels that might be later
1766 // re-used as auto-tuning progresses, or subsequent simulations
1768 PmeGpuProgramStorage pmeGpuProgram;
1769 if (thisRankHasPmeGpuTask)
1772 (deviceStreamManager != nullptr),
1773 "GPU device stream manager should be initialized in order to use GPU for PME.");
1774 GMX_RELEASE_ASSERT((deviceInfo != nullptr),
1775 "GPU device should be initialized in order to use GPU for PME.");
1776 pmeGpuProgram = buildPmeGpuProgram(deviceStreamManager->context());
1779 /* Initiate PME if necessary,
1780 * either on all nodes or on dedicated PME nodes only. */
1781 if (EEL_PME(inputrec->coulombtype) || EVDW_PME(inputrec->vdwtype))
1783 if (mdAtoms && mdAtoms->mdatoms())
1785 nChargePerturbed = mdAtoms->mdatoms()->nChargePerturbed;
1786 if (EVDW_PME(inputrec->vdwtype))
1788 nTypePerturbed = mdAtoms->mdatoms()->nTypePerturbed;
1791 if (cr->npmenodes > 0)
1793 /* The PME only nodes need to know nChargePerturbed(FEP on Q) and nTypePerturbed(FEP on LJ)*/
1794 gmx_bcast(sizeof(nChargePerturbed), &nChargePerturbed, cr->mpi_comm_mysim);
1795 gmx_bcast(sizeof(nTypePerturbed), &nTypePerturbed, cr->mpi_comm_mysim);
1798 if (thisRankHasDuty(cr, DUTY_PME))
1802 // TODO: This should be in the builder.
1803 GMX_RELEASE_ASSERT(!runScheduleWork.simulationWork.useGpuPme
1804 || (deviceStreamManager != nullptr),
1805 "Device stream manager should be valid in order to use GPU "
1808 !runScheduleWork.simulationWork.useGpuPme
1809 || deviceStreamManager->streamIsValid(DeviceStreamType::Pme),
1810 "GPU PME stream should be valid in order to use GPU version of PME.");
1812 const DeviceContext* deviceContext = runScheduleWork.simulationWork.useGpuPme
1813 ? &deviceStreamManager->context()
1815 const DeviceStream* pmeStream =
1816 runScheduleWork.simulationWork.useGpuPme
1817 ? &deviceStreamManager->stream(DeviceStreamType::Pme)
1820 pmedata = gmx_pme_init(cr,
1821 getNumPmeDomains(cr->dd),
1823 nChargePerturbed != 0,
1824 nTypePerturbed != 0,
1825 mdrunOptions.reproducible,
1828 gmx_omp_nthreads_get(emntPME),
1833 pmeGpuProgram.get(),
1836 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
1841 if (EI_DYNAMICS(inputrec->eI))
1843 /* Turn on signal handling on all nodes */
1845 * (A user signal from the PME nodes (if any)
1846 * is communicated to the PP nodes.
1848 signal_handler_install();
1851 pull_t* pull_work = nullptr;
1852 if (thisRankHasDuty(cr, DUTY_PP))
1854 /* Assumes uniform use of the number of OpenMP threads */
1855 walltime_accounting = walltime_accounting_init(gmx_omp_nthreads_get(emntDefault));
1857 if (inputrec->bPull)
1859 /* Initialize pull code */
1860 pull_work = init_pull(fplog,
1861 inputrec->pull.get(),
1866 inputrec->fepvals->init_lambda);
1867 if (inputrec->pull->bXOutAverage || inputrec->pull->bFOutAverage)
1869 initPullHistory(pull_work, &observablesHistory);
1871 if (EI_DYNAMICS(inputrec->eI) && MASTER(cr))
1873 init_pull_output_files(pull_work, filenames.size(), filenames.data(), oenv, startingBehavior);
1877 std::unique_ptr<EnforcedRotation> enforcedRotation;
1880 /* Initialize enforced rotation code */
1881 enforcedRotation = init_rot(fplog,
1894 t_swap* swap = nullptr;
1895 if (inputrec->eSwapCoords != SwapType::No)
1897 /* Initialize ion swapping code */
1898 swap = init_swapcoords(fplog,
1900 opt2fn_master("-swap", filenames.size(), filenames.data(), cr),
1903 &observablesHistory,
1911 /* Let makeConstraints know whether we have essential dynamics constraints. */
1912 auto constr = makeConstraints(
1913 mtop, *inputrec, pull_work, doEssentialDynamics, fplog, cr, ms, &nrnb, wcycle, fr->bMolPBC);
1915 /* Energy terms and groups */
1916 gmx_enerdata_t enerd(mtop.groups.groups[SimulationAtomGroupType::EnergyOutput].size(),
1917 inputrec->fepvals->n_lambda);
1919 // cos acceleration is only supported by md, but older tpr
1920 // files might still combine it with other integrators
1921 GMX_RELEASE_ASSERT(inputrec->cos_accel == 0.0 || inputrec->eI == IntegrationAlgorithm::MD,
1922 "cos_acceleration is only supported by integrator=md");
1924 /* Kinetic energy data */
1925 gmx_ekindata_t ekind(inputrec->opts.ngtc, inputrec->cos_accel, gmx_omp_nthreads_get(emntUpdate));
1927 /* Set up interactive MD (IMD) */
1928 auto imdSession = makeImdSession(inputrec.get(),
1935 MASTER(cr) ? globalState->x : gmx::ArrayRef<gmx::RVec>(),
1939 mdrunOptions.imdOptions,
1942 if (DOMAINDECOMP(cr))
1944 GMX_RELEASE_ASSERT(fr, "fr was NULL while cr->duty was DUTY_PP");
1945 /* This call is not included in init_domain_decomposition mainly
1946 * because fr->cginfo_mb is set later.
1948 dd_init_bondeds(fplog,
1953 domdecOptions.checkBondedInteractions ? DDBondedChecking::All
1954 : DDBondedChecking::ExcludeZeroLimit,
1958 if (runScheduleWork.simulationWork.useGpuBufferOps)
1960 fr->gpuForceReduction[gmx::AtomLocality::Local] = std::make_unique<gmx::GpuForceReduction>(
1961 deviceStreamManager->context(),
1962 deviceStreamManager->stream(gmx::DeviceStreamType::NonBondedLocal),
1964 fr->gpuForceReduction[gmx::AtomLocality::NonLocal] = std::make_unique<gmx::GpuForceReduction>(
1965 deviceStreamManager->context(),
1966 deviceStreamManager->stream(gmx::DeviceStreamType::NonBondedNonLocal),
1970 std::unique_ptr<gmx::StatePropagatorDataGpu> stateGpu;
1971 if (gpusWereDetected
1972 && ((runScheduleWork.simulationWork.useGpuPme && thisRankHasDuty(cr, DUTY_PME))
1973 || runScheduleWork.simulationWork.useGpuBufferOps))
1975 GpuApiCallBehavior transferKind =
1976 (inputrec->eI == IntegrationAlgorithm::MD && !doRerun && !useModularSimulator)
1977 ? GpuApiCallBehavior::Async
1978 : GpuApiCallBehavior::Sync;
1979 GMX_RELEASE_ASSERT(deviceStreamManager != nullptr,
1980 "GPU device stream manager should be initialized to use GPU.");
1981 stateGpu = std::make_unique<gmx::StatePropagatorDataGpu>(
1982 *deviceStreamManager, transferKind, pme_gpu_get_block_size(fr->pmedata), wcycle);
1983 fr->stateGpu = stateGpu.get();
1986 GMX_ASSERT(stopHandlerBuilder_, "Runner must provide StopHandlerBuilder to simulator.");
1987 SimulatorBuilder simulatorBuilder;
1989 simulatorBuilder.add(SimulatorStateData(globalState.get(), &observablesHistory, &enerd, &ekind));
1990 simulatorBuilder.add(std::move(membedHolder));
1991 simulatorBuilder.add(std::move(stopHandlerBuilder_));
1992 simulatorBuilder.add(SimulatorConfig(mdrunOptions, startingBehavior, &runScheduleWork));
1995 simulatorBuilder.add(SimulatorEnv(fplog, cr, ms, mdlog, oenv));
1996 simulatorBuilder.add(Profiling(&nrnb, walltime_accounting, wcycle));
1997 simulatorBuilder.add(ConstraintsParam(
1998 constr.get(), enforcedRotation ? enforcedRotation->getLegacyEnfrot() : nullptr, vsite.get()));
1999 // TODO: Separate `fr` to a separate add, and make the `build` handle the coupling sensibly.
2000 simulatorBuilder.add(LegacyInput(
2001 static_cast<int>(filenames.size()), filenames.data(), inputrec.get(), fr.get()));
2002 simulatorBuilder.add(ReplicaExchangeParameters(replExParams));
2003 simulatorBuilder.add(InteractiveMD(imdSession.get()));
2004 simulatorBuilder.add(SimulatorModules(mdModules_->outputProvider(), mdModules_->notifiers()));
2005 simulatorBuilder.add(CenterOfMassPulling(pull_work));
2006 // Todo move to an MDModule
2007 simulatorBuilder.add(IonSwapping(swap));
2008 simulatorBuilder.add(TopologyData(mtop, mdAtoms.get()));
2009 simulatorBuilder.add(BoxDeformationHandle(deform.get()));
2010 simulatorBuilder.add(std::move(modularSimulatorCheckpointData));
2012 // build and run simulator object based on user-input
2013 auto simulator = simulatorBuilder.build(useModularSimulator);
2016 if (fr->pmePpCommGpu)
2018 // destroy object since it is no longer required. (This needs to be done while the GPU context still exists.)
2019 fr->pmePpCommGpu.reset();
2022 if (inputrec->bPull)
2024 finish_pull(pull_work);
2026 finish_swapcoords(swap);
2030 GMX_RELEASE_ASSERT(pmedata, "pmedata was NULL while cr->duty was not DUTY_PP");
2032 walltime_accounting = walltime_accounting_init(gmx_omp_nthreads_get(emntPME));
2033 gmx_pmeonly(pmedata,
2037 walltime_accounting,
2040 deviceStreamManager.get());
2043 wallcycle_stop(wcycle, ewcRUN);
2045 /* Finish up, write some stuff
2046 * if rerunMD, don't write last frame again
2054 walltime_accounting,
2055 fr ? fr->nbv.get() : nullptr,
2057 EI_DYNAMICS(inputrec->eI) && !isMultiSim(ms));
2059 // clean up cycle counter
2060 wallcycle_destroy(wcycle);
2062 deviceStreamManager.reset(nullptr);
2066 gmx_pme_destroy(pmedata);
2070 // FIXME: this is only here to manually unpin mdAtoms->chargeA_ and state->x,
2071 // before we destroy the GPU context(s)
2072 // Pinned buffers are associated with contexts in CUDA.
2073 // As soon as we destroy GPU contexts after mdrunner() exits, these lines should go.
2074 mdAtoms.reset(nullptr);
2075 globalState.reset(nullptr);
2076 mdModules_.reset(nullptr); // destruct force providers here as they might also use the GPU
2077 gpuBonded.reset(nullptr);
2078 fr.reset(nullptr); // destruct forcerec before gpu
2079 // TODO convert to C++ so we can get rid of these frees
2083 if (!hwinfo_->deviceInfoList.empty())
2085 /* stop the GPU profiler (only CUDA) */
2089 /* With tMPI we need to wait for all ranks to finish deallocation before
2090 * destroying the CUDA context as some tMPI ranks may be sharing
2093 * This is not a concern in OpenCL where we use one context per rank.
2095 * Note: it is safe to not call the barrier on the ranks which do not use GPU,
2096 * but it is easier and more futureproof to call it on the whole node.
2098 * Note that this function needs to be called even if GPUs are not used
2099 * in this run because the PME ranks have no knowledge of whether GPUs
2100 * are used or not, but all ranks need to enter the barrier below.
2101 * \todo Remove this physical node barrier after making sure
2102 * that it's not needed anymore (with a shared GPU run).
2106 physicalNodeComm.barrier();
2109 if (!devFlags.usingCudaAwareMpi)
2111 // Don't reset GPU in case of CUDA-AWARE MPI
2112 // UCX creates CUDA buffers which are cleaned-up as part of MPI_Finalize()
2113 // resetting the device before MPI_Finalize() results in crashes inside UCX
2114 releaseDevice(deviceInfo);
2117 /* Does what it says */
2118 print_date_and_time(fplog, cr->nodeid, "Finished mdrun", gmx_gettime());
2119 walltime_accounting_destroy(walltime_accounting);
2121 // Ensure log file content is written
2124 gmx_fio_flush(logFileHandle);
2127 /* Reset FPEs (important for unit tests) by disabling them. Assumes no
2128 * exceptions were enabled before function was called. */
2131 gmx_fedisableexcept();
2134 auto rc = static_cast<int>(gmx_get_stop_condition());
2137 /* we need to join all threads. The sub-threads join when they
2138 exit this function, but the master thread needs to be told to
2148 Mdrunner::~Mdrunner()
2150 // Clean up of the Manager.
2151 // This will end up getting called on every thread-MPI rank, which is unnecessary,
2152 // but okay as long as threads synchronize some time before adding or accessing
2153 // a new set of restraints.
2154 if (restraintManager_)
2156 restraintManager_->clear();
2157 GMX_ASSERT(restraintManager_->countRestraints() == 0,
2158 "restraints added during runner life time should be cleared at runner "
2163 void Mdrunner::addPotential(std::shared_ptr<gmx::IRestraintPotential> puller, const std::string& name)
2165 GMX_ASSERT(restraintManager_, "Mdrunner must have a restraint manager.");
2166 // Not sure if this should be logged through the md logger or something else,
2167 // but it is helpful to have some sort of INFO level message sent somewhere.
2168 // std::cout << "Registering restraint named " << name << std::endl;
2170 // When multiple restraints are used, it may be wasteful to register them separately.
2171 // Maybe instead register an entire Restraint Manager as a force provider.
2172 restraintManager_->addToSpec(std::move(puller), name);
2175 Mdrunner::Mdrunner(std::unique_ptr<MDModules> mdModules) : mdModules_(std::move(mdModules)) {}
2177 Mdrunner::Mdrunner(Mdrunner&&) noexcept = default;
2179 //NOLINTNEXTLINE(performance-noexcept-move-constructor) working around GCC bug 58265 in CentOS 7
2180 Mdrunner& Mdrunner::operator=(Mdrunner&& /*handle*/) noexcept(BUGFREE_NOEXCEPT_STRING) = default;
2182 class Mdrunner::BuilderImplementation
2185 BuilderImplementation() = delete;
2186 BuilderImplementation(std::unique_ptr<MDModules> mdModules, compat::not_null<SimulationContext*> context);
2187 ~BuilderImplementation();
2189 BuilderImplementation& setExtraMdrunOptions(const MdrunOptions& options,
2190 real forceWarningThreshold,
2191 StartingBehavior startingBehavior);
2193 void addHardwareDetectionResult(const gmx_hw_info_t* hwinfo);
2195 void addDomdec(const DomdecOptions& options);
2197 void addInput(SimulationInputHandle inputHolder);
2199 void addVerletList(int nstlist);
2201 void addReplicaExchange(const ReplicaExchangeParameters& params);
2203 void addNonBonded(const char* nbpu_opt);
2205 void addPME(const char* pme_opt_, const char* pme_fft_opt_);
2207 void addBondedTaskAssignment(const char* bonded_opt);
2209 void addUpdateTaskAssignment(const char* update_opt);
2211 void addHardwareOptions(const gmx_hw_opt_t& hardwareOptions);
2213 void addFilenames(ArrayRef<const t_filenm> filenames);
2215 void addOutputEnvironment(gmx_output_env_t* outputEnvironment);
2217 void addLogFile(t_fileio* logFileHandle);
2219 void addStopHandlerBuilder(std::unique_ptr<StopHandlerBuilder> builder);
2224 // Default parameters copied from runner.h
2225 // \todo Clarify source(s) of default parameters.
2227 const char* nbpu_opt_ = nullptr;
2228 const char* pme_opt_ = nullptr;
2229 const char* pme_fft_opt_ = nullptr;
2230 const char* bonded_opt_ = nullptr;
2231 const char* update_opt_ = nullptr;
2233 MdrunOptions mdrunOptions_;
2235 DomdecOptions domdecOptions_;
2237 ReplicaExchangeParameters replicaExchangeParameters_;
2239 //! Command-line override for the duration of a neighbor list with the Verlet scheme.
2242 //! World communicator, used for hardware detection and task assignment
2243 MPI_Comm libraryWorldCommunicator_ = MPI_COMM_NULL;
2245 //! Multisim communicator handle.
2246 gmx_multisim_t* multiSimulation_;
2248 //! mdrun communicator
2249 MPI_Comm simulationCommunicator_ = MPI_COMM_NULL;
2251 //! Print a warning if any force is larger than this (in kJ/mol nm).
2252 real forceWarningThreshold_ = -1;
2254 //! Whether the simulation will start afresh, or restart with/without appending.
2255 StartingBehavior startingBehavior_ = StartingBehavior::NewSimulation;
2257 //! The modules that comprise the functionality of mdrun.
2258 std::unique_ptr<MDModules> mdModules_;
2260 //! Detected hardware.
2261 const gmx_hw_info_t* hwinfo_ = nullptr;
2263 //! \brief Parallelism information.
2264 gmx_hw_opt_t hardwareOptions_;
2266 //! filename options for simulation.
2267 ArrayRef<const t_filenm> filenames_;
2269 /*! \brief Handle to output environment.
2271 * \todo gmx_output_env_t needs lifetime management.
2273 gmx_output_env_t* outputEnvironment_ = nullptr;
2275 /*! \brief Non-owning handle to MD log file.
2277 * \todo Context should own output facilities for client.
2278 * \todo Improve log file handle management.
2280 * Code managing the FILE* relies on the ability to set it to
2281 * nullptr to check whether the filehandle is valid.
2283 t_fileio* logFileHandle_ = nullptr;
2286 * \brief Builder for simulation stop signal handler.
2288 std::unique_ptr<StopHandlerBuilder> stopHandlerBuilder_ = nullptr;
2291 * \brief Sources for initial simulation state.
2293 * See issue #3652 for near-term refinements to the SimulationInput interface.
2295 * See issue #3379 for broader discussion on API aspects of simulation inputs and outputs.
2297 SimulationInputHandle inputHolder_;
2300 Mdrunner::BuilderImplementation::BuilderImplementation(std::unique_ptr<MDModules> mdModules,
2301 compat::not_null<SimulationContext*> context) :
2302 mdModules_(std::move(mdModules))
2304 libraryWorldCommunicator_ = context->libraryWorldCommunicator_;
2305 simulationCommunicator_ = context->simulationCommunicator_;
2306 multiSimulation_ = context->multiSimulation_.get();
2309 Mdrunner::BuilderImplementation::~BuilderImplementation() = default;
2311 Mdrunner::BuilderImplementation&
2312 Mdrunner::BuilderImplementation::setExtraMdrunOptions(const MdrunOptions& options,
2313 const real forceWarningThreshold,
2314 const StartingBehavior startingBehavior)
2316 mdrunOptions_ = options;
2317 forceWarningThreshold_ = forceWarningThreshold;
2318 startingBehavior_ = startingBehavior;
2322 void Mdrunner::BuilderImplementation::addDomdec(const DomdecOptions& options)
2324 domdecOptions_ = options;
2327 void Mdrunner::BuilderImplementation::addVerletList(int nstlist)
2332 void Mdrunner::BuilderImplementation::addReplicaExchange(const ReplicaExchangeParameters& params)
2334 replicaExchangeParameters_ = params;
2337 Mdrunner Mdrunner::BuilderImplementation::build()
2339 auto newRunner = Mdrunner(std::move(mdModules_));
2341 newRunner.mdrunOptions = mdrunOptions_;
2342 newRunner.pforce = forceWarningThreshold_;
2343 newRunner.startingBehavior = startingBehavior_;
2344 newRunner.domdecOptions = domdecOptions_;
2346 // \todo determine an invariant to check or confirm that all gmx_hw_opt_t objects are valid
2347 newRunner.hw_opt = hardwareOptions_;
2349 // No invariant to check. This parameter exists to optionally override other behavior.
2350 newRunner.nstlist_cmdline = nstlist_;
2352 newRunner.replExParams = replicaExchangeParameters_;
2354 newRunner.filenames = filenames_;
2356 newRunner.libraryWorldCommunicator = libraryWorldCommunicator_;
2358 newRunner.simulationCommunicator = simulationCommunicator_;
2360 // nullptr is a valid value for the multisim handle
2361 newRunner.ms = multiSimulation_;
2365 newRunner.hwinfo_ = hwinfo_;
2369 GMX_THROW(gmx::APIError(
2370 "MdrunnerBuilder::addHardwareDetectionResult() is required before build()"));
2375 newRunner.inputHolder_ = std::move(inputHolder_);
2379 GMX_THROW(gmx::APIError("MdrunnerBuilder::addInput() is required before build()."));
2382 // \todo Clarify ownership and lifetime management for gmx_output_env_t
2383 // \todo Update sanity checking when output environment has clearly specified invariants.
2384 // Initialization and default values for oenv are not well specified in the current version.
2385 if (outputEnvironment_)
2387 newRunner.oenv = outputEnvironment_;
2391 GMX_THROW(gmx::APIError(
2392 "MdrunnerBuilder::addOutputEnvironment() is required before build()"));
2395 newRunner.logFileHandle = logFileHandle_;
2399 newRunner.nbpu_opt = nbpu_opt_;
2403 GMX_THROW(gmx::APIError("MdrunnerBuilder::addNonBonded() is required before build()"));
2406 if (pme_opt_ && pme_fft_opt_)
2408 newRunner.pme_opt = pme_opt_;
2409 newRunner.pme_fft_opt = pme_fft_opt_;
2413 GMX_THROW(gmx::APIError("MdrunnerBuilder::addElectrostatics() is required before build()"));
2418 newRunner.bonded_opt = bonded_opt_;
2422 GMX_THROW(gmx::APIError(
2423 "MdrunnerBuilder::addBondedTaskAssignment() is required before build()"));
2428 newRunner.update_opt = update_opt_;
2432 GMX_THROW(gmx::APIError(
2433 "MdrunnerBuilder::addUpdateTaskAssignment() is required before build() "));
2437 newRunner.restraintManager_ = std::make_unique<gmx::RestraintManager>();
2439 if (stopHandlerBuilder_)
2441 newRunner.stopHandlerBuilder_ = std::move(stopHandlerBuilder_);
2445 newRunner.stopHandlerBuilder_ = std::make_unique<StopHandlerBuilder>();
2451 void Mdrunner::BuilderImplementation::addHardwareDetectionResult(const gmx_hw_info_t* hwinfo)
2456 void Mdrunner::BuilderImplementation::addNonBonded(const char* nbpu_opt)
2458 nbpu_opt_ = nbpu_opt;
2461 void Mdrunner::BuilderImplementation::addPME(const char* pme_opt, const char* pme_fft_opt)
2464 pme_fft_opt_ = pme_fft_opt;
2467 void Mdrunner::BuilderImplementation::addBondedTaskAssignment(const char* bonded_opt)
2469 bonded_opt_ = bonded_opt;
2472 void Mdrunner::BuilderImplementation::addUpdateTaskAssignment(const char* update_opt)
2474 update_opt_ = update_opt;
2477 void Mdrunner::BuilderImplementation::addHardwareOptions(const gmx_hw_opt_t& hardwareOptions)
2479 hardwareOptions_ = hardwareOptions;
2482 void Mdrunner::BuilderImplementation::addFilenames(ArrayRef<const t_filenm> filenames)
2484 filenames_ = filenames;
2487 void Mdrunner::BuilderImplementation::addOutputEnvironment(gmx_output_env_t* outputEnvironment)
2489 outputEnvironment_ = outputEnvironment;
2492 void Mdrunner::BuilderImplementation::addLogFile(t_fileio* logFileHandle)
2494 logFileHandle_ = logFileHandle;
2497 void Mdrunner::BuilderImplementation::addStopHandlerBuilder(std::unique_ptr<StopHandlerBuilder> builder)
2499 stopHandlerBuilder_ = std::move(builder);
2502 void Mdrunner::BuilderImplementation::addInput(SimulationInputHandle inputHolder)
2504 inputHolder_ = std::move(inputHolder);
2507 MdrunnerBuilder::MdrunnerBuilder(std::unique_ptr<MDModules> mdModules,
2508 compat::not_null<SimulationContext*> context) :
2509 impl_{ std::make_unique<Mdrunner::BuilderImplementation>(std::move(mdModules), context) }
2513 MdrunnerBuilder::~MdrunnerBuilder() = default;
2515 MdrunnerBuilder& MdrunnerBuilder::addHardwareDetectionResult(const gmx_hw_info_t* hwinfo)
2517 impl_->addHardwareDetectionResult(hwinfo);
2521 MdrunnerBuilder& MdrunnerBuilder::addSimulationMethod(const MdrunOptions& options,
2522 real forceWarningThreshold,
2523 const StartingBehavior startingBehavior)
2525 impl_->setExtraMdrunOptions(options, forceWarningThreshold, startingBehavior);
2529 MdrunnerBuilder& MdrunnerBuilder::addDomainDecomposition(const DomdecOptions& options)
2531 impl_->addDomdec(options);
2535 MdrunnerBuilder& MdrunnerBuilder::addNeighborList(int nstlist)
2537 impl_->addVerletList(nstlist);
2541 MdrunnerBuilder& MdrunnerBuilder::addReplicaExchange(const ReplicaExchangeParameters& params)
2543 impl_->addReplicaExchange(params);
2547 MdrunnerBuilder& MdrunnerBuilder::addNonBonded(const char* nbpu_opt)
2549 impl_->addNonBonded(nbpu_opt);
2553 MdrunnerBuilder& MdrunnerBuilder::addElectrostatics(const char* pme_opt, const char* pme_fft_opt)
2555 // The builder method may become more general in the future, but in this version,
2556 // parameters for PME electrostatics are both required and the only parameters
2558 if (pme_opt && pme_fft_opt)
2560 impl_->addPME(pme_opt, pme_fft_opt);
2565 gmx::InvalidInputError("addElectrostatics() arguments must be non-null pointers."));
2570 MdrunnerBuilder& MdrunnerBuilder::addBondedTaskAssignment(const char* bonded_opt)
2572 impl_->addBondedTaskAssignment(bonded_opt);
2576 MdrunnerBuilder& MdrunnerBuilder::addUpdateTaskAssignment(const char* update_opt)
2578 impl_->addUpdateTaskAssignment(update_opt);
2582 Mdrunner MdrunnerBuilder::build()
2584 return impl_->build();
2587 MdrunnerBuilder& MdrunnerBuilder::addHardwareOptions(const gmx_hw_opt_t& hardwareOptions)
2589 impl_->addHardwareOptions(hardwareOptions);
2593 MdrunnerBuilder& MdrunnerBuilder::addFilenames(ArrayRef<const t_filenm> filenames)
2595 impl_->addFilenames(filenames);
2599 MdrunnerBuilder& MdrunnerBuilder::addOutputEnvironment(gmx_output_env_t* outputEnvironment)
2601 impl_->addOutputEnvironment(outputEnvironment);
2605 MdrunnerBuilder& MdrunnerBuilder::addLogFile(t_fileio* logFileHandle)
2607 impl_->addLogFile(logFileHandle);
2611 MdrunnerBuilder& MdrunnerBuilder::addStopHandlerBuilder(std::unique_ptr<StopHandlerBuilder> builder)
2613 impl_->addStopHandlerBuilder(std::move(builder));
2617 MdrunnerBuilder& MdrunnerBuilder::addInput(SimulationInputHandle input)
2619 impl_->addInput(std::move(input));
2623 MdrunnerBuilder::MdrunnerBuilder(MdrunnerBuilder&&) noexcept = default;
2625 MdrunnerBuilder& MdrunnerBuilder::operator=(MdrunnerBuilder&&) noexcept = default;