2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 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.
37 * \brief Declaration of low-level functions and fields of GPU state propagator object.
39 * \author Artem Zhmurov <zhmurov@gmail.com>
41 * \ingroup module_mdtypes
43 #ifndef GMX_MDTYPES_STATE_PROPAGATOR_DATA_GPU_IMPL_H
44 #define GMX_MDTYPES_STATE_PROPAGATOR_DATA_GPU_IMPL_H
48 #include "gromacs/gpu_utils/devicebuffer.h"
49 #include "gromacs/math/vectypes.h"
50 #include "gromacs/mdtypes/state_propagator_data_gpu.h"
51 #include "gromacs/utility/classhelpers.h"
56 class StatePropagatorDataGpu::Impl
63 /*! \brief Constructor
65 * The buffers are reallocated only at the reinit call, the padding is
66 * used there for the coordinates buffer. It is needed for PME and added at
67 * the end of the buffer. It is assumed that if the rank has PME duties on the
68 * GPU, all coordinates are copied to the GPU and hence, for this rank, the
69 * coordinates buffer is not split into local and non-local ranges. For other
70 * ranks, the padding size is zero. This works because only one rank ever does
71 * PME work on the GPU, and if that rank also does PP work that is the only
72 * rank. So all coordinates are always transferred.
74 * \note \p commandStream and \p gpuContext are allowed to be nullptr if
75 * StatePropagatorDataGpu is not used in the OpenCL run (e.g. if PME
76 * does not run on the GPU).
78 * \todo Make CommandStream visible in the CPU parts of the code so we
79 * will not have to pass a void*.
80 * \todo Make a Context object visible in CPU parts of the code so we
81 * will not have to pass a void*.
83 * \param[in] commandStream GPU stream, nullptr allowed.
84 * \param[in] gpuContext GPU context, nullptr allowed.
85 * \param[in] transferKind H2D/D2H transfer call behavior (synchronous or not).
86 * \param[in] paddingSize Padding size for coordinates buffer.
88 Impl(const void *commandStream,
89 const void *gpuContext,
90 GpuApiCallBehavior transferKind,
96 /*! \brief Set the ranges for local and non-local atoms and reallocates buffers.
98 * The coordinates buffer is reallocated with the padding added at the end. The
99 * size of padding is set by the constructor.
101 * \param[in] numAtomsLocal Number of atoms in local domain.
102 * \param[in] numAtomsAll Total number of atoms to handle.
104 void reinit(int numAtomsLocal, int numAtomsAll);
106 /*! \brief Returns the range of atoms to be copied based on the copy type (all, local or non-local).
108 * \todo There are at least three versions of the function with this functionality in the code:
109 * this one and two more in NBNXM. These should be unified in a shape of a general function
112 * \param[in] atomLocality If all, local or non-local ranges are needed.
114 * \returns Tuple, containing the index of the first atom in the range and the total number of atoms in the range.
116 std::tuple<int, int> getAtomRangesFromAtomLocality(AtomLocality atomLocality);
119 /*! \brief Get the positions buffer on the GPU.
121 * \returns GPU positions buffer.
123 DeviceBuffer<float> getCoordinates();
125 /*! \brief Copy positions to the GPU memory.
127 * \param[in] h_x Positions in the host memory.
128 * \param[in] atomLocality Locality of the particles to copy.
130 void copyCoordinatesToGpu(gmx::ArrayRef<const gmx::RVec> h_x,
131 AtomLocality atomLocality);
133 /*! \brief Copy positions from the GPU memory.
135 * \param[in] h_x Positions buffer in the host memory.
136 * \param[in] atomLocality Locality of the particles to copy.
138 void copyCoordinatesFromGpu(gmx::ArrayRef<gmx::RVec> h_x,
139 AtomLocality atomLocality);
142 /*! \brief Get the velocities buffer on the GPU.
144 * \returns GPU velocities buffer.
146 DeviceBuffer<float> getVelocities();
148 /*! \brief Copy velocities to the GPU memory.
150 * \param[in] h_v Velocities in the host memory.
151 * \param[in] atomLocality Locality of the particles to copy.
153 void copyVelocitiesToGpu(gmx::ArrayRef<const gmx::RVec> h_v,
154 AtomLocality atomLocality);
156 /*! \brief Copy velocities from the GPU memory.
158 * \param[in] h_v Velocities buffer in the host memory.
159 * \param[in] atomLocality Locality of the particles to copy.
161 void copyVelocitiesFromGpu(gmx::ArrayRef<gmx::RVec> h_v,
162 AtomLocality atomLocality);
165 /*! \brief Get the force buffer on the GPU.
167 * \returns GPU force buffer.
169 DeviceBuffer<float> getForces();
171 /*! \brief Copy forces to the GPU memory.
173 * \param[in] h_f Forces in the host memory.
174 * \param[in] atomLocality Locality of the particles to copy.
176 void copyForcesToGpu(gmx::ArrayRef<const gmx::RVec> h_f,
177 AtomLocality atomLocality);
179 /*! \brief Copy forces from the GPU memory.
181 * \param[in] h_f Forces buffer in the host memory.
182 * \param[in] atomLocality Locality of the particles to copy.
184 void copyForcesFromGpu(gmx::ArrayRef<gmx::RVec> h_f,
185 AtomLocality atomLocality);
187 /*! \brief Synchronize the underlying GPU stream
189 void synchronizeStream();
191 /*! \brief Getter for the number of local atoms.
193 * \returns The number of local atoms.
197 /*! \brief Getter for the total number of atoms.
199 * \returns The total number of atoms.
205 /*! \brief GPU stream.
206 * \todo The stream should be set to non-nullptr once the synchronization points are restored
208 CommandStream commandStream_ = nullptr;
209 /*! \brief GPU context (for OpenCL builds)
210 * \todo Make a Context class usable in CPU code
212 Context gpuContext_ = nullptr;
213 //! Default GPU calls behavior
214 GpuApiCallBehavior transferKind_ = GpuApiCallBehavior::Async;
215 //! Padding size for the coordinates buffer
216 int paddingSize_ = 0;
218 //! Number of local atoms
219 int numAtomsLocal_ = -1;
220 //! Total number of atoms
221 int numAtomsAll_ = -1;
223 //! Device positions buffer
224 DeviceBuffer<float> d_x_;
225 //! Number of particles saved in the positions buffer
227 //! Allocation size for the positions buffer
228 int d_xCapacity_ = -1;
230 //! Device velocities buffer
231 DeviceBuffer<float> d_v_;
232 //! Number of particles saved in the velocities buffer
234 //! Allocation size for the velocities buffer
235 int d_vCapacity_ = -1;
237 //! Device force buffer
238 DeviceBuffer<float> d_f_;
239 //! Number of particles saved in the force buffer
241 //! Allocation size for the force buffer
242 int d_fCapacity_ = -1;
244 /*! \brief Performs the copy of data from host to device buffer.
246 * \todo Template on locality.
248 * \param[in,out] d_data Device-side buffer.
249 * \param[in,out] h_data Host-side buffer.
250 * \param[in] dataSize Device-side data allocation size.
251 * \param[in] atomLocality If all, local or non-local ranges should be copied.
253 void copyToDevice(DeviceBuffer<float> d_data,
254 gmx::ArrayRef<const gmx::RVec> h_data,
256 AtomLocality atomLocality);
258 /*! \brief Performs the copy of data from device to host buffer.
260 * \param[in,out] h_data Host-side buffer.
261 * \param[in,out] d_data Device-side buffer.
262 * \param[in] dataSize Device-side data allocation size.
263 * \param[in] atomLocality If all, local or non-local ranges should be copied.
265 void copyFromDevice(gmx::ArrayRef<gmx::RVec> h_data,
266 DeviceBuffer<float> d_data,
268 AtomLocality atomLocality);
273 #endif // GMX_MDTYPES_STATE_PROPAGATOR_DATA_GPU_IMPL_H