2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 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.
37 * \brief Implements stub GPU 3D FFT routines for CPU-only builds
39 * \author Mark Abraham <mark.j.abraham@gmail.com>
40 * \author Gaurav Garg <gaugarg@nvidia.com>
46 #include "gpu_3dfft.h"
47 #include "gpu_3dfft_impl.h"
50 # include "gpu_3dfft_cufft.h"
52 # include "gpu_3dfft_ocl.h"
54 # include "gpu_3dfft_sycl.h"
56 # include "gpu_3dfft_sycl_rocfft.h"
61 # include "gpu_3dfft_heffte.h"
64 #include "gromacs/utility/arrayref.h"
65 #include "gromacs/utility/exceptions.h"
70 // [[noreturn]] attributes must be added in the common headers, so it's easier to silence the warning here
72 # pragma clang diagnostic push
73 # pragma clang diagnostic ignored "-Wmissing-noreturn"
76 #if (GMX_GPU_CUDA || GMX_GPU_OPENCL || GMX_GPU_SYCL)
78 Gpu3dFft::Gpu3dFft(FftBackend backend,
81 ArrayRef<const int> gridSizesInXForEachRank,
82 ArrayRef<const int> gridSizesInYForEachRank,
84 bool performOutOfPlaceFFT,
85 const DeviceContext& context,
86 const DeviceStream& pmeStream,
88 ivec realGridSizePadded,
89 ivec complexGridSizePadded,
90 DeviceBuffer<float>* realGrid,
91 DeviceBuffer<float>* complexGrid)
96 case FftBackend::Cufft:
97 impl_ = std::make_unique<Gpu3dFft::ImplCuFft>(allocateGrids,
99 gridSizesInXForEachRank,
100 gridSizesInYForEachRank,
102 performOutOfPlaceFFT,
107 complexGridSizePadded,
112 GMX_RELEASE_ASSERT(backend == FftBackend::HeFFTe_CUDA,
113 "Unsupported FFT backend requested");
115 # elif GMX_GPU_OPENCL
118 case FftBackend::Ocl:
119 impl_ = std::make_unique<Gpu3dFft::ImplOcl>(allocateGrids,
121 gridSizesInXForEachRank,
122 gridSizesInYForEachRank,
124 performOutOfPlaceFFT,
129 complexGridSizePadded,
133 default: GMX_THROW(InternalError("Unsupported FFT backend requested"));
138 # if GMX_SYCL_HIPSYCL
139 case FftBackend::SyclRocfft:
140 impl_ = std::make_unique<Gpu3dFft::ImplSyclRocfft>(allocateGrids,
142 gridSizesInXForEachRank,
143 gridSizesInYForEachRank,
145 performOutOfPlaceFFT,
150 complexGridSizePadded,
155 case FftBackend::Sycl:
156 impl_ = std::make_unique<Gpu3dFft::ImplSycl>(allocateGrids,
158 gridSizesInXForEachRank,
159 gridSizesInYForEachRank,
161 performOutOfPlaceFFT,
166 complexGridSizePadded,
170 default: GMX_THROW(InternalError("Unsupported FFT backend requested"));
177 case FftBackend::HeFFTe_CUDA:
180 "HeFFTe_CUDA FFT backend is supported only with GROMACS compiled with CUDA");
181 GMX_RELEASE_ASSERT(heffte::backend::is_enabled<heffte::backend::cufft>::value,
182 "HeFFTe not compiled with CUDA support");
183 impl_ = std::make_unique<Gpu3dFft::ImplHeFfte<heffte::backend::cufft>>(
186 gridSizesInXForEachRank,
187 gridSizesInYForEachRank,
189 performOutOfPlaceFFT,
194 complexGridSizePadded,
199 default: GMX_RELEASE_ASSERT(impl_ != nullptr, "Unsupported FFT backend requested");
206 Gpu3dFft::Gpu3dFft(FftBackend /*backend */,
207 bool /*allocateGrids*/,
209 ArrayRef<const int> /*gridSizesInXForEachRank*/,
210 ArrayRef<const int> /*gridSizesInYForEachRank*/,
212 bool /*performOutOfPlaceFFT*/,
213 const DeviceContext& /*context*/,
214 const DeviceStream& /*pmeStream*/,
215 ivec /*realGridSize*/,
216 ivec /*realGridSizePadded*/,
217 ivec /*complexGridSizePadded*/,
218 DeviceBuffer<float>* /*realGrid*/,
219 DeviceBuffer<float>* /*complexGrid*/)
221 GMX_THROW(InternalError("Cannot run GPU routines in a CPU-only configuration"));
226 Gpu3dFft::~Gpu3dFft() = default;
228 void Gpu3dFft::perform3dFft(gmx_fft_direction dir, CommandEvent* timingEvent)
230 GMX_RELEASE_ASSERT(impl_ != nullptr, "Cannot run GPU routines in a CPU-only configuration");
231 impl_->perform3dFft(dir, timingEvent);
235 # pragma clang diagnostic pop