MatCreateSeqAIJMKL#

Creates a sparse matrix of type MATSEQAIJMKL.

Synopsis#

PetscErrorCode MatCreateSeqAIJMKL(MPI_Comm comm, PetscInt m, PetscInt n, PetscInt nz, const PetscInt nnz[], Mat *A)

Collective

Input Parameters#

  • comm - MPI communicator, set to PETSC_COMM_SELF

  • m - number of rows

  • n - number of columns

  • nz - number of nonzeros per row (same for all rows)

  • nnz - array containing the number of nonzeros in the various rows (possibly different for each row) or NULL

Output Parameter#

  • A - the matrix

Options Database Keys#

  • -mat_aijmkl_no_spmv2 - disable use of the SpMV2 inspector-executor routines

  • -mat_aijmkl_eager_inspection - perform MKL “inspection” phase upon matrix assembly; default is to do “lazy” inspection, performing this step the first time the matrix is applied

Notes#

If nnz is given then nz is ignored

This type inherits from MATSEQAIJ and is largely identical, but uses sparse BLAS routines from Intel MKL whenever possible.

If the installed version of MKL supports the “SpMV2” sparse inspector-executor routines, then those are used by default.

MatMult(), MatMultAdd(), MatMultTranspose(), MatMultTransposeAdd(), MatMatMult(), MatTransposeMatMult(), and MatPtAP() (for symmetric A) operations are currently supported. MKL version 18, update 2 or later is required for MatPtAP(), MatPtAPNumeric() and MatMatMultNumeric().

See Also#

Matrices, Mat, MatCreate(), MatCreateMPIAIJMKL(), MatSetValues()

Level#

intermediate

Location#

src/mat/impls/aij/seq/aijmkl/aijmkl.c

Edit on GitLab

Index of all Mat routines
Table of Contents for all manual pages
Index of all manual pages