cluster_sparse_solver_set_csr

Developer Reference for Intel® oneAPI Math Kernel Library for C

Download PDF

ID 766684

Date 3/22/2024

Version

Public

A newer version of this document is available. Customers should click here to go to the newest version.

Visible to Intel only — GUID: GUID-87778AA2-D959-42E3-B140-8E70D13CB4AF

View Details

cluster_sparse_solver_set_csr_ptrs

Saves internally-provided pointers to the 3-array CSR data corresponding to the specified name.

Syntax

void cluster_sparse_solver_set_csr_ptrs (_MKL_DSS_HANDLE_t pt, _MKL_DSS_EXPORT_DATA name, MKL_INT *rowptr, MKL_INT *colindx, void *vals, const int *comm, MKL_INT *error);

Include Files

mkl_cluster_sparse_solver.h

Description

This routine internally saves the input pointers, rowptr, colindx, and vals, of the 3-array CSR data, which correspond to the provided name. It is assumed that the exported data will be distributed in the same way as the matrix A (as defined by iparm[39]) used in cluster_sparse_solver.

The saved pointers can then be used for exporting corresponding data by means of cluster_sparse_solver_export.

NOTE:

Only call this routine after the factorization phase (phase=22) of the cluster_sparse_solver has been called. Neither pt, nor iparm should be changed after the preceding call to cluster_sparse_solver.

Input Parameters

pt

Array with size of 64.

Handle to internal data structure used in the prior calls to cluster_sparse_solver.

CAUTION:

Do not modify pt after the calls to cluster_sparse_solver.

name

Specifies for which CSR data the pointers are provided.

SPARSE_PTLUQT_L: Factor L from P*A*Q=L*U.
SPARSE_PTLUQT_U: Factor U from P*A*Q=L*U.
SPARSE_DPTLUQT_L: Factor L from P* (D^-1A)*Q=L*U.
SPARSE_DPTLUQT_U: Factor U from P* (D^-1A)*Q=L*U.

rowptr

Array of length at least (local_nrows+1) where local_nrows is the local number of rows, which can be obtained by calling cluster_sparse_solver_get_csr_size. This array contains row indices, such that rowptr[i] - indexing is the first index of row i in the array's vals and colindx. Here, the value of indexing is 0 for zero-based indexing and 1 for one-based indexing, and must be the same as it was for the matrix A used in the preceding calls to cluster_sparse_solver (also stored in iparm[34] ).

Refer to pointerB array description in CSR Format for more details.

colindx

Array of length at least rowptr[local_nrows] – rowptr[0]. Indexing (zero- or one-based) must be the same as for rowptr. For one-based indexing, the array contains the column indices plus one for each non-zero element of the matrix which corresponds to the name. For zero-based indexing, the array contains the column indices for each non-zero element of the matrix.

vals

Array containing non-zero elements of the matrix which corresponds to the name. Its length is equal to length of the colindx array. Refer to values array description in CSR Format for more details.

It will be interpreted internally as float*/double*(MKL_Complex8*/ MKL_Complex16*) depending on mtype (type of the matrix A) and iparm[27] (precision) specified in the preceding call to cluster_sparse_solver.

comm

MPI communicator. The solver uses the Fortran MPI communicator internally. Convert the MPI communicator to Fortran using the MPI_Comm_c2f() function. See the examples in the <install_dir>/examples directory.

Output Parameters

error

The error indicator:

error: Information
0: no error
-1: pt is a null pointer
-2: invalid pt
-3: invalid name
-4: unsupported name
-9: unsupported internal code path, consider switching off non-default iparm parameters for cluster_sparse_solver
-12: internal memory error

NOTE:

Refer to cl_solver_export_c.c for an example using this functionality.

Parent topic: Parallel Direct Sparse Solver for Clusters Interface

Select Your Language

Using Intel.com Search

Quick Links

Recent Searches

Advanced Search

Only search in