2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014,2015,2016 by the GROMACS development team.
5 * Copyright (c) 2017,2018,2019,2020,2021, by the GROMACS development team, led by
6 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
7 * and including many others, as listed in the AUTHORS file in the
8 * top-level source directory and at http://www.gromacs.org.
10 * GROMACS is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU Lesser General Public License
12 * as published by the Free Software Foundation; either version 2.1
13 * of the License, or (at your option) any later version.
15 * GROMACS is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * Lesser General Public License for more details.
20 * You should have received a copy of the GNU Lesser General Public
21 * License along with GROMACS; if not, see
22 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
23 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
25 * If you want to redistribute modifications to GROMACS, please
26 * consider that scientific software is very special. Version
27 * control is crucial - bugs must be traceable. We will be happy to
28 * consider code for inclusion in the official distribution, but
29 * derived work must not be called official GROMACS. Details are found
30 * in the README & COPYING files - if they are missing, get the
31 * official version at http://www.gromacs.org.
33 * To help us fund GROMACS development, we humbly ask that you cite
34 * the research papers on the package. Check out http://www.gromacs.org.
37 * \brief Define CUDA implementation of nbnxn_gpu_data_mgmt.h
39 * \author Szilard Pall <pall.szilard@gmail.com>
48 // TODO We would like to move this down, but the way NbnxmGpu
49 // is currently declared means this has to be before gpu_types.h
50 #include "nbnxm_cuda_types.h"
52 // TODO Remove this comment when the above order issue is resolved
53 #include "gromacs/gpu_utils/cudautils.cuh"
54 #include "gromacs/gpu_utils/device_context.h"
55 #include "gromacs/gpu_utils/gpu_utils.h"
56 #include "gromacs/gpu_utils/gpueventsynchronizer.cuh"
57 #include "gromacs/gpu_utils/pmalloc.h"
58 #include "gromacs/hardware/device_information.h"
59 #include "gromacs/hardware/device_management.h"
60 #include "gromacs/math/vectypes.h"
61 #include "gromacs/mdlib/force_flags.h"
62 #include "gromacs/mdtypes/interaction_const.h"
63 #include "gromacs/mdtypes/md_enums.h"
64 #include "gromacs/nbnxm/atomdata.h"
65 #include "gromacs/nbnxm/gpu_data_mgmt.h"
66 #include "gromacs/nbnxm/gridset.h"
67 #include "gromacs/nbnxm/nbnxm.h"
68 #include "gromacs/nbnxm/nbnxm_gpu.h"
69 #include "gromacs/nbnxm/nbnxm_gpu_data_mgmt.h"
70 #include "gromacs/nbnxm/pairlistsets.h"
71 #include "gromacs/pbcutil/ishift.h"
72 #include "gromacs/timing/gpu_timing.h"
73 #include "gromacs/utility/basedefinitions.h"
74 #include "gromacs/utility/cstringutil.h"
75 #include "gromacs/utility/fatalerror.h"
76 #include "gromacs/utility/real.h"
77 #include "gromacs/utility/smalloc.h"
79 #include "nbnxm_cuda.h"
84 /* This is a heuristically determined parameter for the Kepler
85 * and Maxwell architectures for the minimum size of ci lists by multiplying
86 * this constant with the # of multiprocessors on the current device.
87 * Since the maximum number of blocks per multiprocessor is 16, the ideal
88 * count for small systems is 32 or 48 blocks per multiprocessor. Because
89 * there is a bit of fluctuations in the generated block counts, we use
90 * a target of 44 instead of the ideal value of 48.
92 static unsigned int gpu_min_ci_balanced_factor = 44;
94 void gpu_init_platform_specific(NbnxmGpu* /* nb */)
96 /* set the kernel type for the current GPU */
97 /* pick L1 cache configuration */
98 cuda_set_cacheconfig();
101 void gpu_free_platform_specific(NbnxmGpu* /* nb */)
103 // Nothing specific in CUDA
106 int gpu_min_ci_balanced(NbnxmGpu* nb)
108 return nb != nullptr ? gpu_min_ci_balanced_factor * nb->deviceContext_->deviceInfo().prop.multiProcessorCount
112 void* gpu_get_xq(NbnxmGpu* nb)
116 return static_cast<void*>(nb->atdat->xq);
119 DeviceBuffer<gmx::RVec> gpu_get_f(NbnxmGpu* nb)
123 return reinterpret_cast<DeviceBuffer<gmx::RVec>>(nb->atdat->f);
126 DeviceBuffer<gmx::RVec> gpu_get_fshift(NbnxmGpu* nb)
130 return reinterpret_cast<DeviceBuffer<gmx::RVec>>(nb->atdat->fShift);