2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014,2015,2016,2017,2018,2019, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
36 * \brief Define CUDA implementation of nbnxn_gpu_data_mgmt.h
38 * \author Szilard Pall <pall.szilard@gmail.com>
47 // TODO We would like to move this down, but the way gmx_nbnxn_gpu_t
48 // is currently declared means this has to be before gpu_types.h
49 #include "nbnxm_cuda_types.h"
51 // TODO Remove this comment when the above order issue is resolved
52 #include "gromacs/gpu_utils/cudautils.cuh"
53 #include "gromacs/gpu_utils/gpu_utils.h"
54 #include "gromacs/gpu_utils/pmalloc_cuda.h"
55 #include "gromacs/hardware/gpu_hw_info.h"
56 #include "gromacs/math/vectypes.h"
57 #include "gromacs/mdlib/force_flags.h"
58 #include "gromacs/mdtypes/interaction_const.h"
59 #include "gromacs/mdtypes/md_enums.h"
60 #include "gromacs/nbnxm/atomdata.h"
61 #include "gromacs/nbnxm/gpu_data_mgmt.h"
62 #include "gromacs/nbnxm/nbnxm.h"
63 #include "gromacs/nbnxm/pairlistsets.h"
64 #include "gromacs/pbcutil/ishift.h"
65 #include "gromacs/timing/gpu_timing.h"
66 #include "gromacs/utility/basedefinitions.h"
67 #include "gromacs/utility/cstringutil.h"
68 #include "gromacs/utility/fatalerror.h"
69 #include "gromacs/utility/real.h"
70 #include "gromacs/utility/smalloc.h"
72 #include "nbnxm_cuda.h"
77 /* This is a heuristically determined parameter for the Kepler
78 * and Maxwell architectures for the minimum size of ci lists by multiplying
79 * this constant with the # of multiprocessors on the current device.
80 * Since the maximum number of blocks per multiprocessor is 16, the ideal
81 * count for small systems is 32 or 48 blocks per multiprocessor. Because
82 * there is a bit of fluctuations in the generated block counts, we use
83 * a target of 44 instead of the ideal value of 48.
85 static unsigned int gpu_min_ci_balanced_factor = 44;
88 static void nbnxn_cuda_clear_e_fshift(gmx_nbnxn_cuda_t *nb);
91 static void nbnxn_cuda_free_nbparam_table(cu_nbparam_t *nbparam);
93 /*! \brief Return whether combination rules are used.
95 * \param[in] pointer to nonbonded paramter struct
96 * \return true if combination rules are used in this run, false otherwise
98 static inline bool useLjCombRule(const cu_nbparam_t *nbparam)
100 return (nbparam->vdwtype == evdwCuCUTCOMBGEOM ||
101 nbparam->vdwtype == evdwCuCUTCOMBLB);
104 /*! \brief Initialized the Ewald Coulomb correction GPU table.
106 Tabulates the Ewald Coulomb force and initializes the size/scale
107 and the table GPU array. If called with an already allocated table,
108 it just re-uploads the table.
110 static void init_ewald_coulomb_force_table(const interaction_const_t *ic,
113 if (nbp->coulomb_tab != nullptr)
115 nbnxn_cuda_free_nbparam_table(nbp);
118 nbp->coulomb_tab_scale = ic->tabq_scale;
119 initParamLookupTable(nbp->coulomb_tab, nbp->coulomb_tab_texobj,
120 ic->tabq_coul_F, ic->tabq_size);
124 /*! Initializes the atomdata structure first time, it only gets filled at
126 static void init_atomdata_first(cu_atomdata_t *ad, int ntypes)
131 stat = cudaMalloc((void**)&ad->shift_vec, SHIFTS*sizeof(*ad->shift_vec));
132 CU_RET_ERR(stat, "cudaMalloc failed on ad->shift_vec");
133 ad->bShiftVecUploaded = false;
135 stat = cudaMalloc((void**)&ad->fshift, SHIFTS*sizeof(*ad->fshift));
136 CU_RET_ERR(stat, "cudaMalloc failed on ad->fshift");
138 stat = cudaMalloc((void**)&ad->e_lj, sizeof(*ad->e_lj));
139 CU_RET_ERR(stat, "cudaMalloc failed on ad->e_lj");
140 stat = cudaMalloc((void**)&ad->e_el, sizeof(*ad->e_el));
141 CU_RET_ERR(stat, "cudaMalloc failed on ad->e_el");
143 /* initialize to nullptr poiters to data that is not allocated here and will
144 need reallocation in nbnxn_cuda_init_atomdata */
148 /* size -1 indicates that the respective array hasn't been initialized yet */
153 /*! Selects the Ewald kernel type, analytical on SM 3.0 and later, tabulated on
154 earlier GPUs, single or twin cut-off. */
155 static int pick_ewald_kernel_type(bool bTwinCut)
157 bool bUseAnalyticalEwald, bForceAnalyticalEwald, bForceTabulatedEwald;
160 /* Benchmarking/development environment variables to force the use of
161 analytical or tabulated Ewald kernel. */
162 bForceAnalyticalEwald = (getenv("GMX_CUDA_NB_ANA_EWALD") != nullptr);
163 bForceTabulatedEwald = (getenv("GMX_CUDA_NB_TAB_EWALD") != nullptr);
165 if (bForceAnalyticalEwald && bForceTabulatedEwald)
167 gmx_incons("Both analytical and tabulated Ewald CUDA non-bonded kernels "
168 "requested through environment variables.");
171 /* By default use analytical Ewald. */
172 bUseAnalyticalEwald = true;
173 if (bForceAnalyticalEwald)
177 fprintf(debug, "Using analytical Ewald CUDA kernels\n");
180 else if (bForceTabulatedEwald)
182 bUseAnalyticalEwald = false;
186 fprintf(debug, "Using tabulated Ewald CUDA kernels\n");
190 /* Use twin cut-off kernels if requested by bTwinCut or the env. var.
191 forces it (use it for debugging/benchmarking only). */
192 if (!bTwinCut && (getenv("GMX_CUDA_NB_EWALD_TWINCUT") == nullptr))
194 kernel_type = bUseAnalyticalEwald ? eelCuEWALD_ANA : eelCuEWALD_TAB;
198 kernel_type = bUseAnalyticalEwald ? eelCuEWALD_ANA_TWIN : eelCuEWALD_TAB_TWIN;
204 /*! Copies all parameters related to the cut-off from ic to nbp */
205 static void set_cutoff_parameters(cu_nbparam_t *nbp,
206 const interaction_const_t *ic,
207 const PairlistParams &listParams)
209 nbp->ewald_beta = ic->ewaldcoeff_q;
210 nbp->sh_ewald = ic->sh_ewald;
211 nbp->epsfac = ic->epsfac;
212 nbp->two_k_rf = 2.0 * ic->k_rf;
213 nbp->c_rf = ic->c_rf;
214 nbp->rvdw_sq = ic->rvdw * ic->rvdw;
215 nbp->rcoulomb_sq = ic->rcoulomb * ic->rcoulomb;
216 nbp->rlistOuter_sq = listParams.rlistOuter * listParams.rlistOuter;
217 nbp->rlistInner_sq = listParams.rlistInner * listParams.rlistInner;
218 nbp->useDynamicPruning = listParams.useDynamicPruning;
220 nbp->sh_lj_ewald = ic->sh_lj_ewald;
221 nbp->ewaldcoeff_lj = ic->ewaldcoeff_lj;
223 nbp->rvdw_switch = ic->rvdw_switch;
224 nbp->dispersion_shift = ic->dispersion_shift;
225 nbp->repulsion_shift = ic->repulsion_shift;
226 nbp->vdw_switch = ic->vdw_switch;
229 /*! Initializes the nonbonded parameter data structure. */
230 static void init_nbparam(cu_nbparam_t *nbp,
231 const interaction_const_t *ic,
232 const PairlistParams &listParams,
233 const nbnxn_atomdata_t::Params &nbatParams)
237 ntypes = nbatParams.numTypes;
239 set_cutoff_parameters(nbp, ic, listParams);
241 /* The kernel code supports LJ combination rules (geometric and LB) for
242 * all kernel types, but we only generate useful combination rule kernels.
243 * We currently only use LJ combination rule (geometric and LB) kernels
244 * for plain cut-off LJ. On Maxwell the force only kernels speed up 15%
245 * with PME and 20% with RF, the other kernels speed up about half as much.
246 * For LJ force-switch the geometric rule would give 7% speed-up, but this
247 * combination is rarely used. LJ force-switch with LB rule is more common,
248 * but gives only 1% speed-up.
250 if (ic->vdwtype == evdwCUT)
252 switch (ic->vdw_modifier)
255 case eintmodPOTSHIFT:
256 switch (nbatParams.comb_rule)
259 nbp->vdwtype = evdwCuCUT;
262 nbp->vdwtype = evdwCuCUTCOMBGEOM;
265 nbp->vdwtype = evdwCuCUTCOMBLB;
268 gmx_incons("The requested LJ combination rule is not implemented in the CUDA GPU accelerated kernels!");
271 case eintmodFORCESWITCH:
272 nbp->vdwtype = evdwCuFSWITCH;
274 case eintmodPOTSWITCH:
275 nbp->vdwtype = evdwCuPSWITCH;
278 gmx_incons("The requested VdW interaction modifier is not implemented in the CUDA GPU accelerated kernels!");
281 else if (ic->vdwtype == evdwPME)
283 if (ic->ljpme_comb_rule == ljcrGEOM)
285 assert(nbatParams.comb_rule == ljcrGEOM);
286 nbp->vdwtype = evdwCuEWALDGEOM;
290 assert(nbatParams.comb_rule == ljcrLB);
291 nbp->vdwtype = evdwCuEWALDLB;
296 gmx_incons("The requested VdW type is not implemented in the CUDA GPU accelerated kernels!");
299 if (ic->eeltype == eelCUT)
301 nbp->eeltype = eelCuCUT;
303 else if (EEL_RF(ic->eeltype))
305 nbp->eeltype = eelCuRF;
307 else if ((EEL_PME(ic->eeltype) || ic->eeltype == eelEWALD))
309 /* Initially rcoulomb == rvdw, so it's surely not twin cut-off. */
310 nbp->eeltype = pick_ewald_kernel_type(false);
314 /* Shouldn't happen, as this is checked when choosing Verlet-scheme */
315 gmx_incons("The requested electrostatics type is not implemented in the CUDA GPU accelerated kernels!");
318 /* generate table for PME */
319 nbp->coulomb_tab = nullptr;
320 if (nbp->eeltype == eelCuEWALD_TAB || nbp->eeltype == eelCuEWALD_TAB_TWIN)
322 init_ewald_coulomb_force_table(ic, nbp);
325 /* set up LJ parameter lookup table */
326 if (!useLjCombRule(nbp))
328 initParamLookupTable(nbp->nbfp, nbp->nbfp_texobj,
329 nbatParams.nbfp.data(), 2*ntypes*ntypes);
332 /* set up LJ-PME parameter lookup table */
333 if (ic->vdwtype == evdwPME)
335 initParamLookupTable(nbp->nbfp_comb, nbp->nbfp_comb_texobj,
336 nbatParams.nbfp_comb.data(), 2*ntypes);
340 /*! Re-generate the GPU Ewald force table, resets rlist, and update the
341 * electrostatic type switching to twin cut-off (or back) if needed. */
342 void gpu_pme_loadbal_update_param(const nonbonded_verlet_t *nbv,
343 const interaction_const_t *ic)
345 if (!nbv || !nbv->useGpu())
349 cu_nbparam_t *nbp = nbv->gpu_nbv->nbparam;
351 set_cutoff_parameters(nbp, ic, nbv->pairlistSets().params());
353 nbp->eeltype = pick_ewald_kernel_type(ic->rcoulomb != ic->rvdw);
355 init_ewald_coulomb_force_table(ic, nbp);
358 /*! Initializes the pair list data structure. */
359 static void init_plist(cu_plist_t *pl)
361 /* initialize to nullptr pointers to data that is not allocated here and will
362 need reallocation in nbnxn_gpu_init_pairlist */
368 /* size -1 indicates that the respective array hasn't been initialized yet */
375 pl->imask_nalloc = -1;
377 pl->excl_nalloc = -1;
378 pl->haveFreshList = false;
381 /*! Initializes the timings data structure. */
382 static void init_timings(gmx_wallclock_gpu_nbnxn_t *t)
391 for (i = 0; i < 2; i++)
393 for (j = 0; j < 2; j++)
395 t->ktime[i][j].t = 0.0;
396 t->ktime[i][j].c = 0;
400 t->pruneTime.t = 0.0;
401 t->dynamicPruneTime.c = 0;
402 t->dynamicPruneTime.t = 0.0;
405 /*! Initializes simulation constant data. */
406 static void cuda_init_const(gmx_nbnxn_cuda_t *nb,
407 const interaction_const_t *ic,
408 const PairlistParams &listParams,
409 const nbnxn_atomdata_t::Params &nbatParams)
411 init_atomdata_first(nb->atdat, nbatParams.numTypes);
412 init_nbparam(nb->nbparam, ic, listParams, nbatParams);
414 /* clear energy and shift force outputs */
415 nbnxn_cuda_clear_e_fshift(nb);
419 gpu_init(const gmx_device_info_t *deviceInfo,
420 const interaction_const_t *ic,
421 const PairlistParams &listParams,
422 const nbnxn_atomdata_t *nbat,
424 gmx_bool bLocalAndNonlocal)
428 gmx_nbnxn_cuda_t *nb;
431 snew(nb->nbparam, 1);
432 snew(nb->plist[InteractionLocality::Local], 1);
433 if (bLocalAndNonlocal)
435 snew(nb->plist[InteractionLocality::NonLocal], 1);
438 nb->bUseTwoStreams = bLocalAndNonlocal;
440 nb->timers = new cu_timers_t();
441 snew(nb->timings, 1);
444 pmalloc((void**)&nb->nbst.e_lj, sizeof(*nb->nbst.e_lj));
445 pmalloc((void**)&nb->nbst.e_el, sizeof(*nb->nbst.e_el));
446 pmalloc((void**)&nb->nbst.fshift, SHIFTS * sizeof(*nb->nbst.fshift));
448 init_plist(nb->plist[InteractionLocality::Local]);
450 /* set device info, just point it to the right GPU among the detected ones */
451 nb->dev_info = deviceInfo;
453 /* local/non-local GPU streams */
454 stat = cudaStreamCreate(&nb->stream[InteractionLocality::Local]);
455 CU_RET_ERR(stat, "cudaStreamCreate on stream[InterationLocality::Local] failed");
456 if (nb->bUseTwoStreams)
458 init_plist(nb->plist[InteractionLocality::NonLocal]);
460 /* Note that the device we're running on does not have to support
461 * priorities, because we are querying the priority range which in this
462 * case will be a single value.
464 int highest_priority;
465 stat = cudaDeviceGetStreamPriorityRange(nullptr, &highest_priority);
466 CU_RET_ERR(stat, "cudaDeviceGetStreamPriorityRange failed");
468 stat = cudaStreamCreateWithPriority(&nb->stream[InteractionLocality::NonLocal],
471 CU_RET_ERR(stat, "cudaStreamCreateWithPriority on stream[InteractionLocality::NonLocal] failed");
474 /* init events for sychronization (timing disabled for performance reasons!) */
475 stat = cudaEventCreateWithFlags(&nb->nonlocal_done, cudaEventDisableTiming);
476 CU_RET_ERR(stat, "cudaEventCreate on nonlocal_done failed");
477 stat = cudaEventCreateWithFlags(&nb->misc_ops_and_local_H2D_done, cudaEventDisableTiming);
478 CU_RET_ERR(stat, "cudaEventCreate on misc_ops_and_local_H2D_done failed");
480 /* WARNING: CUDA timings are incorrect with multiple streams.
481 * This is the main reason why they are disabled by default.
483 // TODO: Consider turning on by default when we can detect nr of streams.
484 nb->bDoTime = (getenv("GMX_ENABLE_GPU_TIMING") != nullptr);
488 init_timings(nb->timings);
491 /* set the kernel type for the current GPU */
492 /* pick L1 cache configuration */
493 cuda_set_cacheconfig();
495 cuda_init_const(nb, ic, listParams, nbat->params());
499 fprintf(debug, "Initialized CUDA data structures.\n");
505 void gpu_init_pairlist(gmx_nbnxn_cuda_t *nb,
506 const NbnxnPairlistGpu *h_plist,
507 const InteractionLocality iloc)
510 bool bDoTime = (nb->bDoTime && !h_plist->sci.empty());
511 cudaStream_t stream = nb->stream[iloc];
512 cu_plist_t *d_plist = nb->plist[iloc];
514 if (d_plist->na_c < 0)
516 d_plist->na_c = h_plist->na_ci;
520 if (d_plist->na_c != h_plist->na_ci)
522 sprintf(sbuf, "In cu_init_plist: the #atoms per cell has changed (from %d to %d)",
523 d_plist->na_c, h_plist->na_ci);
528 gpu_timers_t::Interaction &iTimers = nb->timers->interaction[iloc];
532 iTimers.pl_h2d.openTimingRegion(stream);
533 iTimers.didPairlistH2D = true;
536 Context context = nullptr;
538 reallocateDeviceBuffer(&d_plist->sci, h_plist->sci.size(),
539 &d_plist->nsci, &d_plist->sci_nalloc, context);
540 copyToDeviceBuffer(&d_plist->sci, h_plist->sci.data(), 0, h_plist->sci.size(),
541 stream, GpuApiCallBehavior::Async,
542 bDoTime ? iTimers.pl_h2d.fetchNextEvent() : nullptr);
544 reallocateDeviceBuffer(&d_plist->cj4, h_plist->cj4.size(),
545 &d_plist->ncj4, &d_plist->cj4_nalloc, context);
546 copyToDeviceBuffer(&d_plist->cj4, h_plist->cj4.data(), 0, h_plist->cj4.size(),
547 stream, GpuApiCallBehavior::Async,
548 bDoTime ? iTimers.pl_h2d.fetchNextEvent() : nullptr);
550 reallocateDeviceBuffer(&d_plist->imask, h_plist->cj4.size()*c_nbnxnGpuClusterpairSplit,
551 &d_plist->nimask, &d_plist->imask_nalloc, context);
553 reallocateDeviceBuffer(&d_plist->excl, h_plist->excl.size(),
554 &d_plist->nexcl, &d_plist->excl_nalloc, context);
555 copyToDeviceBuffer(&d_plist->excl, h_plist->excl.data(), 0, h_plist->excl.size(),
556 stream, GpuApiCallBehavior::Async,
557 bDoTime ? iTimers.pl_h2d.fetchNextEvent() : nullptr);
561 iTimers.pl_h2d.closeTimingRegion(stream);
564 /* the next use of thist list we be the first one, so we need to prune */
565 d_plist->haveFreshList = true;
568 void gpu_upload_shiftvec(gmx_nbnxn_cuda_t *nb,
569 const nbnxn_atomdata_t *nbatom)
571 cu_atomdata_t *adat = nb->atdat;
572 cudaStream_t ls = nb->stream[InteractionLocality::Local];
574 /* only if we have a dynamic box */
575 if (nbatom->bDynamicBox || !adat->bShiftVecUploaded)
577 cu_copy_H2D_async(adat->shift_vec, nbatom->shift_vec.data(),
578 SHIFTS * sizeof(*adat->shift_vec), ls);
579 adat->bShiftVecUploaded = true;
583 /*! Clears the first natoms_clear elements of the GPU nonbonded force output array. */
584 static void nbnxn_cuda_clear_f(gmx_nbnxn_cuda_t *nb, int natoms_clear)
587 cu_atomdata_t *adat = nb->atdat;
588 cudaStream_t ls = nb->stream[InteractionLocality::Local];
590 stat = cudaMemsetAsync(adat->f, 0, natoms_clear * sizeof(*adat->f), ls);
591 CU_RET_ERR(stat, "cudaMemsetAsync on f falied");
594 /*! Clears nonbonded shift force output array and energy outputs on the GPU. */
595 static void nbnxn_cuda_clear_e_fshift(gmx_nbnxn_cuda_t *nb)
598 cu_atomdata_t *adat = nb->atdat;
599 cudaStream_t ls = nb->stream[InteractionLocality::Local];
601 stat = cudaMemsetAsync(adat->fshift, 0, SHIFTS * sizeof(*adat->fshift), ls);
602 CU_RET_ERR(stat, "cudaMemsetAsync on fshift falied");
603 stat = cudaMemsetAsync(adat->e_lj, 0, sizeof(*adat->e_lj), ls);
604 CU_RET_ERR(stat, "cudaMemsetAsync on e_lj falied");
605 stat = cudaMemsetAsync(adat->e_el, 0, sizeof(*adat->e_el), ls);
606 CU_RET_ERR(stat, "cudaMemsetAsync on e_el falied");
609 void gpu_clear_outputs(gmx_nbnxn_cuda_t *nb, int flags)
611 nbnxn_cuda_clear_f(nb, nb->atdat->natoms);
612 /* clear shift force array and energies if the outputs were
613 used in the current step */
614 if (flags & GMX_FORCE_VIRIAL)
616 nbnxn_cuda_clear_e_fshift(nb);
620 void gpu_init_atomdata(gmx_nbnxn_cuda_t *nb,
621 const nbnxn_atomdata_t *nbat)
626 bool bDoTime = nb->bDoTime;
627 cu_timers_t *timers = nb->timers;
628 cu_atomdata_t *d_atdat = nb->atdat;
629 cudaStream_t ls = nb->stream[InteractionLocality::Local];
631 natoms = nbat->numAtoms();
636 /* time async copy */
637 timers->atdat.openTimingRegion(ls);
640 /* need to reallocate if we have to copy more atoms than the amount of space
641 available and only allocate if we haven't initialized yet, i.e d_atdat->natoms == -1 */
642 if (natoms > d_atdat->nalloc)
644 nalloc = over_alloc_small(natoms);
646 /* free up first if the arrays have already been initialized */
647 if (d_atdat->nalloc != -1)
649 freeDeviceBuffer(&d_atdat->f);
650 freeDeviceBuffer(&d_atdat->xq);
651 freeDeviceBuffer(&d_atdat->atom_types);
652 freeDeviceBuffer(&d_atdat->lj_comb);
655 stat = cudaMalloc((void **)&d_atdat->f, nalloc*sizeof(*d_atdat->f));
656 CU_RET_ERR(stat, "cudaMalloc failed on d_atdat->f");
657 stat = cudaMalloc((void **)&d_atdat->xq, nalloc*sizeof(*d_atdat->xq));
658 CU_RET_ERR(stat, "cudaMalloc failed on d_atdat->xq");
659 if (useLjCombRule(nb->nbparam))
661 stat = cudaMalloc((void **)&d_atdat->lj_comb, nalloc*sizeof(*d_atdat->lj_comb));
662 CU_RET_ERR(stat, "cudaMalloc failed on d_atdat->lj_comb");
666 stat = cudaMalloc((void **)&d_atdat->atom_types, nalloc*sizeof(*d_atdat->atom_types));
667 CU_RET_ERR(stat, "cudaMalloc failed on d_atdat->atom_types");
670 d_atdat->nalloc = nalloc;
674 d_atdat->natoms = natoms;
675 d_atdat->natoms_local = nbat->natoms_local;
677 /* need to clear GPU f output if realloc happened */
680 nbnxn_cuda_clear_f(nb, nalloc);
683 if (useLjCombRule(nb->nbparam))
685 cu_copy_H2D_async(d_atdat->lj_comb, nbat->params().lj_comb.data(),
686 natoms*sizeof(*d_atdat->lj_comb), ls);
690 cu_copy_H2D_async(d_atdat->atom_types, nbat->params().type.data(),
691 natoms*sizeof(*d_atdat->atom_types), ls);
696 timers->atdat.closeTimingRegion(ls);
700 static void nbnxn_cuda_free_nbparam_table(cu_nbparam_t *nbparam)
702 if (nbparam->eeltype == eelCuEWALD_TAB || nbparam->eeltype == eelCuEWALD_TAB_TWIN)
704 destroyParamLookupTable(nbparam->coulomb_tab, nbparam->coulomb_tab_texobj);
708 void gpu_free(gmx_nbnxn_cuda_t *nb)
711 cu_atomdata_t *atdat;
712 cu_nbparam_t *nbparam;
720 nbparam = nb->nbparam;
722 nbnxn_cuda_free_nbparam_table(nbparam);
724 stat = cudaEventDestroy(nb->nonlocal_done);
725 CU_RET_ERR(stat, "cudaEventDestroy failed on timers->nonlocal_done");
726 stat = cudaEventDestroy(nb->misc_ops_and_local_H2D_done);
727 CU_RET_ERR(stat, "cudaEventDestroy failed on timers->misc_ops_and_local_H2D_done");
732 /* The non-local counters/stream (second in the array) are needed only with DD. */
733 for (int i = 0; i <= (nb->bUseTwoStreams ? 1 : 0); i++)
735 stat = cudaStreamDestroy(nb->stream[i]);
736 CU_RET_ERR(stat, "cudaStreamDestroy failed on stream");
740 if (!useLjCombRule(nb->nbparam))
742 destroyParamLookupTable(nbparam->nbfp, nbparam->nbfp_texobj);
746 if (nbparam->vdwtype == evdwCuEWALDGEOM || nbparam->vdwtype == evdwCuEWALDLB)
748 destroyParamLookupTable(nbparam->nbfp_comb, nbparam->nbfp_comb_texobj);
751 stat = cudaFree(atdat->shift_vec);
752 CU_RET_ERR(stat, "cudaFree failed on atdat->shift_vec");
753 stat = cudaFree(atdat->fshift);
754 CU_RET_ERR(stat, "cudaFree failed on atdat->fshift");
756 stat = cudaFree(atdat->e_lj);
757 CU_RET_ERR(stat, "cudaFree failed on atdat->e_lj");
758 stat = cudaFree(atdat->e_el);
759 CU_RET_ERR(stat, "cudaFree failed on atdat->e_el");
761 freeDeviceBuffer(&atdat->f);
762 freeDeviceBuffer(&atdat->xq);
763 freeDeviceBuffer(&atdat->atom_types);
764 freeDeviceBuffer(&atdat->lj_comb);
767 auto *plist = nb->plist[InteractionLocality::Local];
768 freeDeviceBuffer(&plist->sci);
769 freeDeviceBuffer(&plist->cj4);
770 freeDeviceBuffer(&plist->imask);
771 freeDeviceBuffer(&plist->excl);
773 if (nb->bUseTwoStreams)
775 auto *plist_nl = nb->plist[InteractionLocality::NonLocal];
776 freeDeviceBuffer(&plist_nl->sci);
777 freeDeviceBuffer(&plist_nl->cj4);
778 freeDeviceBuffer(&plist_nl->imask);
779 freeDeviceBuffer(&plist_nl->excl);
784 pfree(nb->nbst.e_lj);
785 nb->nbst.e_lj = nullptr;
787 pfree(nb->nbst.e_el);
788 nb->nbst.e_el = nullptr;
790 pfree(nb->nbst.fshift);
791 nb->nbst.fshift = nullptr;
800 fprintf(debug, "Cleaned up CUDA data structures.\n");
804 //! This function is documented in the header file
805 gmx_wallclock_gpu_nbnxn_t *gpu_get_timings(gmx_nbnxn_cuda_t *nb)
807 return (nb != nullptr && nb->bDoTime) ? nb->timings : nullptr;
810 void gpu_reset_timings(nonbonded_verlet_t* nbv)
812 if (nbv->gpu_nbv && nbv->gpu_nbv->bDoTime)
814 init_timings(nbv->gpu_nbv->timings);
818 int gpu_min_ci_balanced(gmx_nbnxn_cuda_t *nb)
820 return nb != nullptr ?
821 gpu_min_ci_balanced_factor*nb->dev_info->prop.multiProcessorCount : 0;
825 gmx_bool gpu_is_kernel_ewald_analytical(const gmx_nbnxn_cuda_t *nb)
827 return ((nb->nbparam->eeltype == eelCuEWALD_ANA) ||
828 (nb->nbparam->eeltype == eelCuEWALD_ANA_TWIN));
831 void *gpu_get_command_stream(gmx_nbnxn_gpu_t *nb,
832 const InteractionLocality iloc)
836 return static_cast<void *>(&nb->stream[iloc]);
839 void *gpu_get_xq(gmx_nbnxn_gpu_t *nb)
843 return static_cast<void *>(nb->atdat->xq);
846 void *gpu_get_f(gmx_nbnxn_gpu_t *nb)
850 return static_cast<void *>(nb->atdat->f);
853 rvec *gpu_get_fshift(gmx_nbnxn_gpu_t *nb)
857 return reinterpret_cast<rvec *>(nb->atdat->fshift);