KokkosSparse::spgemm_numeric¶

Defined in header KokkosSparse_spgemm.hpp

Note

The minimal header KokkosSparse_spgemm_numeric.hpp is also sufficient for version 1 below.

template <typename KernelHandle, typename alno_row_view_t_, typename alno_nnz_view_t_, typename ascalar_nnz_view_t_,
          typename blno_row_view_t_, typename blno_nnz_view_t_, typename bscalar_nnz_view_t_, typename clno_row_view_t_,
          typename clno_nnz_view_t_, typename cscalar_nnz_view_t_>
void spgemm_numeric(KernelHandle *handle, typename KernelHandle::const_nnz_lno_t m,
                    typename KernelHandle::const_nnz_lno_t n, typename KernelHandle::const_nnz_lno_t k,
                    alno_row_view_t_ row_mapA, alno_nnz_view_t_ entriesA, ascalar_nnz_view_t_ valuesA,
                    bool transposeA, blno_row_view_t_ row_mapB, blno_nnz_view_t_ entriesB, bscalar_nnz_view_t_ valuesB,
                    bool transposeB, clno_row_view_t_ row_mapC, clno_nnz_view_t_ &entriesC,
                    cscalar_nnz_view_t_ &valuesC,
                    typename KernelHandle::const_nnz_lno_t block_dim = 1);

template <class KernelHandle, class AMatrix, class BMatrix, class CMatrix>
void spgemm_numeric(KernelHandle& kh, const AMatrix& A, const bool Amode, const BMatrix& B, const bool Bmode,
                    CMatrix& C);

Performs the numeric phase of the multiplication of two sparse matrices.

The symbolic phase (spgemm_symbolic) must have previously been called with the same handle and input matrices (except for scalar values).

\[C = A*B\]

Computes the column indices and values of matrix C. Also computes the row map of C if it was not previously computed by spgemm_symbolic.
Same as 1. but uses CrsMatrix objects for a more compact interface.

Warning

While transpose flags are part of the interface, the algorithm is only implemented for non-transposed A and B.

spgemm_numeric can reuse the analysis performed by spgemm_symbolic, as long as the sparsity patterns (row map and entries) of A and B do not change. If only the values of A and B have changed, then spgemm_numeric can always be called with the same handle to recompute the values of C.

Parameters¶

Note

For all parameters that are shared with spgemm_symbolic (dimensions, row maps, entries and matrices), the parameters to spgemm_numeric must match what was previously passed into spgemm_symbolic with the same handle. For example, it is not valid to call spgemm_symbolic with one version of entriesA, and then call spgemm_numeric with a different entriesA but the same handle.

handle:: spadd kernels handle obtained from an instance of KokkosKernels::KokkosKernelsHandle. Must be the same handle that was previously passed into spgemm_symbolic with the same input matrices.
m, n, k:: dimensions of the matrices. m is the number of rows in A and C. n is the number of columns in A and the number of rows in B. k is the number of columns in B and C.
row_mapA, row_mapB:: row maps of the input matrices A and B.
entriesA, entriesB:: column indices of the entries in each row of A and B. In most cases, these should be in sorted and merged order (see note in spgemm_symbolic).
valuesA, valuesB:: values of the entries in each row of A and B.
transposeA, transposeB:: transposition operator to be applied to row_mapA, entriesA and row_mapB, entriesB respectively. Must both be false; the algorithm is not yet implemented for other cases.
row_mapC:: row map of the output matrix C.
entriesC:: column indices of the entries in each row of C. Must be allocated (though not initialized) by the user to size handle->get_spgemm_handle()->get_c_nnz().
valuesC:: values of the entries in each row of C. Must be allocated (though not initialized) by the user to size handle->get_spgemm_handle()->get_c_nnz().
Amode, Bmode:: transposition operation to apply to A and B respectively. Must both be false; the algorithm is not yet implemented for other cases.
A, B, C:: three instances of KokkosSparse::CrsMatrix. A and B are inputs parameters only. The output is written into the values of C. In most cases, A and B should be in sorted and merged order (see note in spgemm_symbolic).

Type Requirements¶

alno_row_view_t_, alno_nnz_view_t_ and ascalar_nnz_view_t_ must be compatible with the types expected by KernelHandle
- std::is_same_v<typename KernelHandle::const_size_type, typename alno_row_view_t_::const_value_type> == true
- std::is_same_v<typename KernelHandle::const_nnz_lno_t, typename alno_nnz_view_t_::const_value_type> == true
- std::is_same_v<typename KernelHandle::const_nnz_scalar_t, typename ascalar_nnz_view_t_::const_value_type> == true
blno_row_view_t_, blno_nnz_view_t_ and bscalar_nnz_view_t_ must be compatible with the types expected by KernelHandle
- std::is_same_v<typename KernelHandle::const_size_type, typename blno_row_view_t_::const_value_type> == true
- std::is_same_v<typename KernelHandle::const_nnz_lno_t, typename blno_nnz_view_t_::const_value_type> == true
- std::is_same_v<typename KernelHandle::const_nnz_scalar_t, typename bscalar_nnz_view_t_::const_value_type> == true
clno_row_view_t_, clno_nnz_view_t_ and cscalar_nnz_view_t_ must be compatible with the type expected by KernelHandle and clno_nnz_view_t_ and cscalar_nnz_view_t_ must be non-const.
- std::is_same_v<typename KernelHandle::const_size_type, typename clno_row_view_t_::const_value_type> == true
- std::is_same_v<typename KernelHandle::const_nnz_lno_t, typename clno_nnz_view_t_::const_value_type> == true
- std::is_same_v<typename KernelHandle::const_nnz_scalar_t, typename cscalar_nnz_view_t_::const_value_type> == true
- std::is_same_v<typename KernelHandle::const_nnz_lno_t, typename clno_nnz_view_t_::const_value_type> == true
- std::is_same_v<typename cscalar_nnz_view_t_::value_type, typename cscalar_nnz_view_t_::non_const_value_type> == true

Example¶

#include "Kokkos_Core.hpp"

#include "KokkosKernels_default_types.hpp"
#include "KokkosSparse_spgemm.hpp"

#include "KokkosKernels_Test_Structured_Matrix.hpp"

using Scalar  = default_scalar;
using Ordinal = default_lno_t;
using Offset  = default_size_type;
using Layout  = default_layout;

int main(int argc, char* argv[]) {
  Kokkos::initialize();

  using device_type = typename Kokkos::Device<
      Kokkos::DefaultExecutionSpace,
      typename Kokkos::DefaultExecutionSpace::memory_space>;
  using execution_space = typename device_type::execution_space;
  using memory_space    = typename device_type::memory_space;
  using matrix_type =
      typename KokkosSparse::CrsMatrix<Scalar, Ordinal, device_type, void,
                                       Offset>;

  int return_value = 0;

  {
    // The mat_structure view is used to generate a matrix using
    // finite difference (FD) or finite element (FE) discretization
    // on a cartesian grid.
    // Each row corresponds to an axis (x, y and z)
    // In each row the first entry is the number of grid point in
    // that direction, the second and third entries are used to apply
    // BCs in that direction.
    Kokkos::View<Ordinal* [3], Kokkos::HostSpace> mat_structure(
        "Matrix Structure", 2);
    mat_structure(0, 0) = 10;  // Request 10 grid point in 'x' direction
    mat_structure(0, 1) = 1;   // Add BC to the left
    mat_structure(0, 2) = 1;   // Add BC to the right
    mat_structure(1, 0) = 10;  // Request 10 grid point in 'y' direction
    mat_structure(1, 1) = 1;   // Add BC to the bottom
    mat_structure(1, 2) = 1;   // Add BC to the top

    matrix_type A =
        Test::generate_structured_matrix2D<matrix_type>("FD", mat_structure);
    matrix_type B =
        Test::generate_structured_matrix2D<matrix_type>("FE", mat_structure);
    matrix_type C;

    // Create KokkosKernelHandle
    using KernelHandle = KokkosKernels::Experimental::KokkosKernelsHandle<
        Offset, Ordinal, Scalar, execution_space, memory_space, memory_space>;
    KernelHandle kh;
    kh.set_team_work_size(16);
    kh.set_dynamic_scheduling(true);

    // Select an spgemm algorithm, limited by configuration at compile-time and
    // set via the handle Some options: {SPGEMM_KK_MEMORY, SPGEMM_KK_SPEED,
    // SPGEMM_KK_MEMSPEED, /*SPGEMM_CUSPARSE, */ SPGEMM_MKL}
    std::string myalg("SPGEMM_KK_MEMORY");
    KokkosSparse::SPGEMMAlgorithm spgemm_algorithm =
        KokkosSparse::StringToSPGEMMAlgorithm(myalg);
    kh.create_spgemm_handle(spgemm_algorithm);

    KokkosSparse::spgemm_symbolic(kh, A, false, B, false, C);
    KokkosSparse::spgemm_numeric(kh, A, false, B, false, C);

    std::cout << "spgemm was performed correctly!" << std::endl;
  }

  Kokkos::finalize();

  return return_value;
}