openclBISAI.hpp

/*
  Copyright 2022 Equinor ASA

  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_OPENCLBISAI_HPP
#define OPM_OPENCLBISAI_HPP

#include <mutex>

#include <opm/simulators/linalg/gpubridge/opencl/opencl.hpp>
#include <opm/simulators/linalg/gpubridge/opencl/openclBILU0.hpp>
#include <opm/simulators/linalg/gpubridge/opencl/openclPreconditioner.hpp>

namespace Opm::Accelerator {

template<class Scalar> class BlockedMatrix;

template<class Scalar, unsigned int block_size>
class openclBISAI : public openclPreconditioner<Scalar,block_size>
{
    using Base = openclPreconditioner<Scalar,block_size>;

    using Base::N;
    using Base::Nb;
    using Base::nnz;
    using Base::nnzb;
    using Base::verbosity;
    using Base::context;
    using Base::queue;
    using Base::events;
    using Base::err;

private:
    std::once_flag initialize;

    std::vector<int> colPointers;
    std::vector<int> rowIndices;
    std::vector<int> diagIndex;
    std::vector<int> csrToCscOffsetMap;
    std::vector<Scalar> invLvals;
    std::vector<Scalar> invUvals;

    cl::Buffer d_colPointers;
    cl::Buffer d_rowIndices;
    cl::Buffer d_csrToCscOffsetMap;
    cl::Buffer d_diagIndex;
    cl::Buffer d_LUvals;
    cl::Buffer d_invDiagVals;
    cl::Buffer d_invLvals;
    cl::Buffer d_invUvals;
    cl::Buffer d_invL_x;

    bool opencl_ilu_parallel;
    std::unique_ptr<openclBILU0<Scalar,block_size>> bilu0;

    struct subsystemStructure {
        std::vector<int> subsystemPointers;
        std::vector<int> nzIndices;
        std::vector<int> knownRhsIndices;
        std::vector<int> unknownRhsIndices;
    };

    struct subsystemStructureGPU {
        cl::Buffer subsystemPointers;
        cl::Buffer nzIndices;
        cl::Buffer knownRhsIndices;
        cl::Buffer unknownRhsIndices;
    } ;

    subsystemStructure lower, upper;
    subsystemStructureGPU d_lower, d_upper;

    void buildLowerSubsystemsStructures();

    void buildUpperSubsystemsStructures();

public:
    openclBISAI(bool opencl_ilu_parallel, int verbosity);

    // set own Opencl variables, but also that of the bilu0 preconditioner
    void setOpencl(std::shared_ptr<cl::Context>& context,
                   std::shared_ptr<cl::CommandQueue>& queue) override;

    // analysis, extract parallelism
    bool analyze_matrix(BlockedMatrix<Scalar>* mat) override;
    bool analyze_matrix(BlockedMatrix<Scalar>* mat,
                        BlockedMatrix<Scalar>* jacMat) override;

    // ilu_decomposition
    bool create_preconditioner(BlockedMatrix<Scalar>* mat) override;
    bool create_preconditioner(BlockedMatrix<Scalar>* mat,
                               BlockedMatrix<Scalar>* jacMat) override;

    // apply preconditioner, x = prec(y)
    void apply(const cl::Buffer& y,
               cl::Buffer& x,
               WellContributions<Scalar>& wellContribs) override;
};

std::vector<int> buildCsrToCscOffsetMap(std::vector<int> colPointers, std::vector<int> rowIndices);

} // namespace Opm::Accelerator

#endif