2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2016,2017,2018,2019,2020, 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 * Describes common routines and types for PME tests.
39 * \author Aleksei Iupinov <a.yupinov@gmail.com>
40 * \ingroup module_ewald
42 #ifndef GMX_EWALD_PME_TEST_COMMON_H
43 #define GMX_EWALD_PME_TEST_COMMON_H
49 #include "gromacs/ewald/pme.h"
50 #include "gromacs/ewald/pme_gpu_internal.h"
51 #include "gromacs/math/gmxcomplex.h"
52 #include "gromacs/mdtypes/state_propagator_data_gpu.h"
53 #include "gromacs/utility/unique_cptr.h"
55 #include "testutils/test_device.h"
66 //! Hardware code path being tested
67 enum class CodePath : int
73 //! Total number of code paths
77 //! Return a string useful for human-readable messages describing a \c codePath.
78 const char* codePathToString(CodePath codePath);
80 // Convenience typedefs
81 //! A safe pointer type for PME.
82 typedef gmx::unique_cptr<gmx_pme_t, gmx_pme_destroy> PmeSafePointer;
84 typedef ArrayRef<const real> ChargesVector;
86 typedef std::vector<RVec> CoordinatesVector;
88 typedef ArrayRef<RVec> ForcesVector;
90 typedef ArrayRef<const IVec> GridLineIndicesVector;
91 /*! \brief Spline parameters (theta or dtheta).
92 * A reference to a single dimension's spline data; this means (atomCount * pmeOrder) values or derivatives.
94 typedef ArrayRef<const real> SplineParamsDimVector;
95 /*! \brief Spline parameters (theta or dtheta) in all 3 dimensions
97 typedef std::array<SplineParamsDimVector, DIM> SplineParamsVector;
99 //! Non-zero grid values for test input; keys are 3d indices (IVec)
100 template<typename ValueType>
101 using SparseGridValuesInput = std::map<IVec, ValueType>;
102 //! Non-zero real grid values
103 typedef SparseGridValuesInput<real> SparseRealGridValuesInput;
104 //! Non-zero complex grid values
105 typedef SparseGridValuesInput<t_complex> SparseComplexGridValuesInput;
106 //! Non-zero grid values for test output; keys are string representations of the cells' 3d indices (IVec); this allows for better sorting.
107 template<typename ValueType>
108 using SparseGridValuesOutput = std::map<std::string, ValueType>;
109 //! Non-zero real grid values
110 typedef SparseGridValuesOutput<real> SparseRealGridValuesOutput;
111 //! Non-zero complex grid values
112 typedef SparseGridValuesOutput<t_complex> SparseComplexGridValuesOutput;
113 //! TODO: make proper C++ matrix for the whole Gromacs, get rid of this
114 typedef std::array<real, DIM * DIM> Matrix3x3;
116 enum class PmeSolveAlgorithm : int
118 //! Coulomb electrostatics
122 //! Total number of solvers
128 //! Tells if this generally valid PME input is supported for this mode
129 bool pmeSupportsInputForMode(const gmx_hw_info_t& hwinfo, const t_inputrec* inputRec, CodePath mode);
131 //! Spline moduli are computed in double precision, so they're very good in single precision
132 constexpr int64_t c_splineModuliSinglePrecisionUlps = 1;
133 /*! \brief For double precision checks, the recursive interpolation
134 * and use of trig functions in make_dft_mod require a lot more flops,
135 * and thus opportunity for deviation between implementations. */
136 uint64_t getSplineModuliDoublePrecisionUlps(int splineOrder);
140 //! PME initialization
141 PmeSafePointer pmeInitWrapper(const t_inputrec* inputRec,
143 const DeviceContext* deviceContext,
144 const DeviceStream* deviceStream,
145 const PmeGpuProgram* pmeGpuProgram,
146 const Matrix3x3& box,
147 real ewaldCoeff_q = 1.0F,
148 real ewaldCoeff_lj = 1.0F);
150 //! Simple PME initialization based on inputrec only
151 PmeSafePointer pmeInitEmpty(const t_inputrec* inputRec);
153 //! Make a GPU state-propagator manager
154 std::unique_ptr<StatePropagatorDataGpu> makeStatePropagatorDataGpu(const gmx_pme_t& pme,
155 const DeviceContext* deviceContext,
156 const DeviceStream* deviceStream);
157 //! PME initialization with atom data and system box
158 void pmeInitAtoms(gmx_pme_t* pme,
159 StatePropagatorDataGpu* stateGpu,
161 const CoordinatesVector& coordinates,
162 const ChargesVector& charges);
163 //! PME spline computation and charge spreading
164 void pmePerformSplineAndSpread(gmx_pme_t* pme, CodePath mode, bool computeSplines, bool spreadCharges);
166 void pmePerformSolve(const gmx_pme_t* pme,
168 PmeSolveAlgorithm method,
170 GridOrdering gridOrdering,
171 bool computeEnergyAndVirial);
172 //! PME force gathering
173 void pmePerformGather(gmx_pme_t* pme,
175 ForcesVector& forces); //NOLINT(google-runtime-references)
176 //! PME test finalization before fetching the outputs
177 void pmeFinalizeTest(const gmx_pme_t* pme, CodePath mode);
181 //! Setting atom spline values or derivatives to be used in spread/gather
182 void pmeSetSplineData(const gmx_pme_t* pme,
184 const SplineParamsDimVector& splineValues,
185 PmeSplineDataType type,
188 //! Setting gridline indices be used in spread/gather
189 void pmeSetGridLineIndices(gmx_pme_t* pme, CodePath mode, const GridLineIndicesVector& gridLineIndices);
190 //! Setting real grid to be used in gather
191 void pmeSetRealGrid(const gmx_pme_t* pme, CodePath mode, const SparseRealGridValuesInput& gridValues);
192 void pmeSetComplexGrid(const gmx_pme_t* pme,
194 GridOrdering gridOrdering,
195 const SparseComplexGridValuesInput& gridValues);
199 //! Getting the single dimension's spline values or derivatives
200 SplineParamsDimVector pmeGetSplineData(const gmx_pme_t* pme, CodePath mode, PmeSplineDataType type, int dimIndex);
201 //! Getting the gridline indices
202 GridLineIndicesVector pmeGetGridlineIndices(const gmx_pme_t* pme, CodePath mode);
203 //! Getting the real grid (spreading output of pmePerformSplineAndSpread())
204 SparseRealGridValuesOutput pmeGetRealGrid(const gmx_pme_t* pme, CodePath mode);
205 //! Getting the complex grid output of pmePerformSolve()
206 SparseComplexGridValuesOutput pmeGetComplexGrid(const gmx_pme_t* pme, CodePath mode, GridOrdering gridOrdering);
207 //! Getting the reciprocal energy and virial
208 PmeOutput pmeGetReciprocalEnergyAndVirial(const gmx_pme_t* pme, CodePath mode, PmeSolveAlgorithm method);
210 struct PmeTestHardwareContext
212 //! Hardware path for the code being tested.
214 //! Returns a human-readable context description line
215 std::string description() const;
216 //! Pointer to the global test hardware device (if on GPU)
217 TestDevice* testDevice_ = nullptr;
218 //! PME GPU program if needed
219 PmeGpuProgramStorage pmeGpuProgram_ = nullptr;
220 // Constructor for CPU context
221 PmeTestHardwareContext();
222 // Constructor for GPU context
223 explicit PmeTestHardwareContext(TestDevice* testDevice);
225 //! Get the code path
226 CodePath codePath() const { return codePath_; }
227 //! Get the PME GPU program
228 const PmeGpuProgram* pmeGpuProgram() const
230 return codePath() == CodePath::GPU ? pmeGpuProgram_.get() : nullptr;
233 const DeviceContext* deviceContext() const
235 return codePath() == CodePath::GPU ? &testDevice_->deviceContext() : nullptr;
238 const DeviceStream* deviceStream() const
240 return codePath() == CodePath::GPU ? &testDevice_->deviceStream() : nullptr;
243 //! Activate the context (set the device)
244 void activate() const;
247 std::vector<std::unique_ptr<PmeTestHardwareContext>> createPmeTestHardwareContextList();
252 #endif // GMX_EWALD_PME_TEST_COMMON_H