2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2017,2018,2019,2020,2021, 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 Implements common internal types for different NBNXN GPU implementations
38 * \author Szilárd Páll <pall.szilard@gmail.com>
39 * \ingroup module_nbnxm
42 #ifndef GMX_MDLIB_NBNXN_GPU_COMMON_TYPES_H
43 #define GMX_MDLIB_NBNXN_GPU_COMMON_TYPES_H
47 #include "gromacs/mdtypes/interaction_const.h"
48 #include "gromacs/mdtypes/locality.h"
49 #include "gromacs/utility/enumerationhelpers.h"
55 # include "gromacs/gpu_utils/gpuregiontimer_ocl.h"
59 # include "gromacs/gpu_utils/gpuregiontimer.cuh"
63 # include "gromacs/gpu_utils/gpuregiontimer_sycl.h"
66 /*! \brief Macro definining default for the prune kernel's j4 processing concurrency.
68 * The GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY macro allows compile-time override with the default value of 4.
70 #ifndef GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY
71 # define GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY 4
73 //! Default for the prune kernel's j4 processing concurrency.
74 static constexpr int c_pruneKernelJ4Concurrency = GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY;
77 * \brief Staging area for temporary data downloaded from the GPU.
79 * Since SYCL buffers already have host-side storage, this is a bit redundant.
80 * But it allows prefetching of the data from GPU, and brings GPU backends closer together.
86 //! electrostatic energy
87 float* eElec = nullptr;
89 Float3* fShift = nullptr;
93 * \brief Nonbonded atom data - both inputs and outputs.
99 //! number of local atoms
101 //! allocation size for the atom data (xq, f)
104 //! atom coordinates + charges, size \ref numAtoms
105 DeviceBuffer<Float4> xq;
106 //! force output array, size \ref numAtoms
107 DeviceBuffer<Float3> f;
109 //! LJ energy output, size 1
110 DeviceBuffer<float> eLJ;
111 //! Electrostatics energy input, size 1
112 DeviceBuffer<float> eElec;
115 DeviceBuffer<Float3> fShift;
117 //! number of atom types
119 //! atom type indices, size \ref numAtoms
120 DeviceBuffer<int> atomTypes;
121 //! sqrt(c6),sqrt(c12) size \ref numAtoms
122 DeviceBuffer<Float2> ljComb;
125 DeviceBuffer<Float3> shiftVec;
126 //! true if the shift vector has been uploaded
127 bool shiftVecUploaded;
131 * \brief Parameters required for the GPU nonbonded calculations.
136 //! type of electrostatics
137 enum Nbnxm::ElecType elecType;
138 //! type of VdW impl.
139 enum Nbnxm::VdwType vdwType;
141 //! charge multiplication factor
143 //! Reaction-field/plain cutoff electrostatics const.
145 //! Reaction-field electrostatics constant
147 //! Ewald/PME parameter
149 //! Ewald/PME correction term subtracted from the direct-space potential
151 //! LJ-Ewald/PME correction term added to the correction potential
153 //! LJ-Ewald/PME coefficient
156 //! Coulomb cut-off squared
159 //! VdW cut-off squared
161 //! VdW switched cut-off
163 //! Full, outer pair-list cut-off squared
165 //! Inner, dynamic pruned pair-list cut-off squared
167 //! True if we use dynamic pair-list pruning
168 bool useDynamicPruning;
170 //! VdW shift dispersion constants
171 shift_consts_t dispersion_shift;
172 //! VdW shift repulsion constants
173 shift_consts_t repulsion_shift;
174 //! VdW switch constants
175 switch_consts_t vdw_switch;
177 /* LJ non-bonded parameters - accessed through texture memory */
178 //! nonbonded parameter table with C6/C12 pairs per atom type-pair, 2*ntype^2 elements
179 DeviceBuffer<float> nbfp;
180 //! texture object bound to nbfp
181 DeviceTexture nbfp_texobj;
182 //! nonbonded parameter table per atom type, 2*ntype elements
183 DeviceBuffer<float> nbfp_comb;
184 //! texture object bound to nbfp_comb
185 DeviceTexture nbfp_comb_texobj;
187 /* Ewald Coulomb force table data - accessed through texture memory */
188 //! table scale/spacing
189 float coulomb_tab_scale;
190 //! pointer to the table in the device memory
191 DeviceBuffer<float> coulomb_tab;
192 //! texture object bound to coulomb_tab
193 DeviceTexture coulomb_tab_texobj;
199 using gmx::AtomLocality;
200 using gmx::InteractionLocality;
203 * \brief GPU region timers used for timing GPU kernels and H2D/D2H transfers.
205 * The two-sized arrays hold the local and non-local values and should always
206 * be indexed with eintLocal/eintNonlocal.
211 * \brief Timers for local or non-local coordinate/force transfers
215 //! timer for x/q H2D transfers (l/nl, every step)
216 GpuRegionTimer nb_h2d;
217 //! timer for f D2H transfer (l/nl, every step)
218 GpuRegionTimer nb_d2h;
222 * \brief Timers for local or non-local interaction related operations
226 //! timer for pair-list H2D transfers (l/nl, every PS step)
227 GpuRegionTimer pl_h2d;
228 //! true when a pair-list transfer has been done at this step
229 bool didPairlistH2D = false;
230 //! timer for non-bonded kernels (l/nl, every step)
232 //! timer for the 1st pass list pruning kernel (l/nl, every PS step)
233 GpuRegionTimer prune_k;
234 //! true when we timed pruning and the timings need to be accounted for
235 bool didPrune = false;
236 //! timer for rolling pruning kernels (l/nl, frequency depends on chunk size)
237 GpuRegionTimer rollingPrune_k;
238 //! true when we timed rolling pruning (at the previous step) and the timings need to be accounted for
239 bool didRollingPrune = false;
242 //! timer for atom data transfer (every PS step)
243 GpuRegionTimer atdat;
244 //! timers for coordinate/force transfers (every step)
245 gmx::EnumerationArray<AtomLocality, XFTransfers> xf;
246 //! timers for interaction related transfers
247 gmx::EnumerationArray<InteractionLocality, Nbnxm::GpuTimers::Interaction> interaction;
251 * \brief GPU pair list structure */
254 //! number of atoms per cluster
257 //! size of sci, # of i clusters in the list
259 //! allocation size of sci
261 //! list of i-cluster ("super-clusters")
262 DeviceBuffer<nbnxn_sci_t> sci;
264 //! total # of 4*j clusters
266 //! allocation size of cj4
268 //! 4*j cluster list, contains j cluster number and index into the i cluster list
269 DeviceBuffer<nbnxn_cj4_t> cj4;
270 //! # of 4*j clusters * # of warps
272 //! allocation size of imask
274 //! imask for 2 warps for each 4*j cluster group
275 DeviceBuffer<unsigned int> imask;
276 //! atom interaction bits
277 DeviceBuffer<nbnxn_excl_t> excl;
280 //! allocation size of excl
283 /* parameter+variables for normal and rolling pruning */
284 //! true after search, indicates that initial pruning with outer pruning is needed
286 //! the number of parts/steps over which one cycle of rolling pruning takes places
287 int rollingPruningNumParts;
288 //! the next part to which the rolling pruning needs to be applied
289 int rollingPruningPart;