DILUKernels.hpp

/*
  Copyright 2024 SINTEF AS

  This file is part of the Open Porous Media project (OPM).

  OPM is free software: you can redistribute it and/or modify
  it under the terms of the GNU General Public License as published by
  the Free Software Foundation, either version 3 of the License, or
  (at your option) any later version.

  OPM is distributed in the hope that it will be useful,
  but WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  GNU General Public License for more details.

  You should have received a copy of the GNU General Public License
  along with OPM.  If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef OPM_DILU_KERNELS_HPP
#define OPM_DILU_KERNELS_HPP

#include <cstddef>
#include <cuda.h>
#include <cuda_runtime.h>
#include <opm/simulators/linalg/gpuistl/detail/kernel_enums.hpp>
#include <vector>

namespace Opm::gpuistl::detail::DILU
{

template <class T, int blocksize>
void solveLowerLevelSet(T* reorderedMat,
                        int* rowIndices,
                        int* colIndices,
                        int* indexConversion,
                        int startIdx,
                        int rowsInLevelSet,
                        const T* dInv,
                        const T* d,
                        T* v,
                        int threadBlockSize,
                        cudaStream_t stream);

template <class T, int blocksize>
void solveLowerLevelSetNoReorder(const T* mat,
                                 const int* rowIndices,
                                 const int* colIndices,
                                 const size_t* indexConversion,
                                 int startIdx,
                                 int rowsInLevelSet,
                                 const T* dInv,
                                 const T* d,
                                 T* v,
                                 int threadBlockSize,
                                 cudaStream_t stream);

template <int blocksize, class LinearSolverScalar, class MatrixScalar, class DiagonalScalar>
void solveLowerLevelSetSplit(MatrixScalar* reorderedUpperMat,
                             int* rowIndices,
                             int* colIndices,
                             int* indexConversion,
                             int startIdx,
                             int rowsInLevelSet,
                             const DiagonalScalar* dInv,
                             const LinearSolverScalar* d,
                             LinearSolverScalar* v,
                             int threadBlockSize,
                             cudaStream_t stream);

template <class T, int blocksize>
void solveUpperLevelSet(T* reorderedMat,
                        int* rowIndices,
                        int* colIndices,
                        int* indexConversion,
                        int startIdx,
                        int rowsInLevelSet,
                        const T* dInv,
                        T* v,
                        int threadBlockSize,
                        cudaStream_t stream);

template <class T, int blocksize>
void solveUpperLevelSetNoReorder(const T* mat,
                                 const int* rowIndices,
                                 const int* colIndices,
                                 const size_t* indexConversion,
                                 int startIdx,
                                 int rowsInLevelSet,
                                 const T* dInv,
                                 T* v,
                                 int threadBlockSize,
                                 cudaStream_t stream);

template <int blocksize, class LinearSolverScalar, class MatrixScalar, class DiagonalScalar>
void solveUpperLevelSetSplit(MatrixScalar* reorderedUpperMat,
                             int* rowIndices,
                             int* colIndices,
                             int* indexConversion,
                             int startIdx,
                             int rowsInLevelSet,
                             const DiagonalScalar* dInv,
                             LinearSolverScalar* v,
                             int threadBlockSize,
                             cudaStream_t stream);

template <class T, int blocksize>
void computeDiluDiagonal(T* reorderedMat,
                         int* rowIndices,
                         int* colIndices,
                         int* reorderedToNatural,
                         int* naturalToReordered,
                         size_t* diagIndices,
                         int startIdx,
                         int rowsInLevelSet,
                         T* dInv,
                         int threadBlockSize);

template <class T, int blocksize>
void computeDiluDiagonalNoReorder(const T* mat,
                                  const int* rowIndices,
                                  const int* colIndices,
                                  const size_t* indexConversion,
                                  const size_t* diagIndices,
                                  int startIdx,
                                  int rowsInLevelSet,
                                  T* dInv,
                                  int threadBlockSize);

template <int blocksize, class InputScalar, class OutputScalar, MatrixStorageMPScheme>
void computeDiluDiagonalSplit(const InputScalar* srcReorderedLowerMat,
                              int* lowerRowIndices,
                              int* lowerColIndices,
                              const InputScalar* srcReorderedUpperMat,
                              int* upperRowIndices,
                              int* upperColIndices,
                              const InputScalar* srcDiagonal,
                              int* reorderedToNatural,
                              int* naturalToReordered,
                              int startIdx,
                              int rowsInLevelSet,
                              InputScalar* dInv,
                              OutputScalar* dstDiagonal,
                              OutputScalar* dstLowerMat,
                              OutputScalar* dstUpperMat,
                              int threadBlockSize);

} // namespace Opm::gpuistl::detail::DILU
#endif