Class CudaSparseContext

Wrapper class for cusparseContext. Provides all fundamental API functions as methods.

Inheritance

System.Object

CudaSparseContext

Implements

System.IDisposable

Inherited Members

System.Object.Equals(System.Object)

System.Object.Equals(System.Object, System.Object)

System.Object.GetHashCode()

System.Object.GetType()

System.Object.MemberwiseClone()

System.Object.ReferenceEquals(System.Object, System.Object)

System.Object.ToString()

Namespace: ManagedCuda.CudaSparse

Assembly: CudaSparse.dll

Syntax

public class CudaSparseContext : IDisposable

Constructors

CudaSparseContext()

Creates a new CudaSparseContext

Declaration

public CudaSparseContext()

CudaSparseContext(CUstream)

Creates a new CudaSparseContext and sets the cuda stream to use

Declaration

public CudaSparseContext(CUstream stream)

Parameters

Type	Name	Description
CUstream	stream	A valid CUDA stream created with cudaStreamCreate() (or 0 for the default stream)

Properties

Handle

Returns the wrapped cusparseContext handle

Declaration

public cusparseContext Handle { get; }

Property Value

Type	Description
cusparseContext

Methods

Axpyi(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Axpyi(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Axpyi(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Axpyi(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Axpyi(cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Axpyi(cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Axpyi(Double, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(double alpha, CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Axpyi(Single, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, cusparseIndexBase)

Addition of a scalar multiple of a sparse vector x and a dense vector y

Declaration

public void Axpyi(float alpha, CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Bsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Declaration

public void Bsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A. The number of rows of sparse matrix C is m(= mb*blockDim).
System.Int32	nb	number of block columns of sparse matrix A. The number of columns of sparse matrix C is n(= nb*blockDim).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb*blockDim² non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the non-zero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<cuDoubleComplex>	csrValC	array of nnz (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the non-zero elements of matrix C.

Bsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Declaration

public void Bsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A. The number of rows of sparse matrix C is m(= mb*blockDim).
System.Int32	nb	number of block columns of sparse matrix A. The number of columns of sparse matrix C is n(= nb*blockDim).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb*blockDim² non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the non-zero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<cuFloatComplex>	csrValC	array of nnz (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the non-zero elements of matrix C.

Bsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Declaration

public void Bsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A. The number of rows of sparse matrix C is m(= mb*blockDim).
System.Int32	nb	number of block columns of sparse matrix A. The number of columns of sparse matrix C is n(= nb*blockDim).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb*blockDim² non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the non-zero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<System.Double>	csrValC	array of nnz (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the non-zero elements of matrix C.

Bsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Declaration

public void Bsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A. The number of rows of sparse matrix C is m(= mb*blockDim).
System.Int32	nb	number of block columns of sparse matrix A. The number of columns of sparse matrix C is n(= nb*blockDim).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb*blockDim² non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the non-zero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<System.Single>	csrValC	array of nnz (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the non-zero elements of matrix C.

Bsric02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Bsric02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsric02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Bsric02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsric02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Bsric02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsric02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Bsric02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsric02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Bsric02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsric02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsric02ZeroPivot(CudaSparseBsric02Info, CudaDeviceVariable<Int32>)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXbsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Bsric02ZeroPivot(CudaSparseBsric02Info info, CudaDeviceVariable<int> position)

Parameters

Type	Name	Description
CudaSparseBsric02Info	info	info contains structural zero or numerical zero if the user already called bsrsv2_analysis() or bsrsv2_solve().
CudaDeviceVariable<System.Int32>	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Bsric02ZeroPivot(CudaSparseBsric02Info, ref Int32)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXbsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Bsric02ZeroPivot(CudaSparseBsric02Info info, ref int position)

Parameters

Type	Name	Description
CudaSparseBsric02Info	info	info contains structural zero or numerical zero if the user already called bsrsv2_analysis() or bsrsv2_solve().
System.Int32	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Bsrilu02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Bsrilu02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Bsrilu02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Bsrilu02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Bsrilu02(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA_ValM, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA_ValM	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with fillin and no pivoting: A = LU

Declaration

public void Bsrilu02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with fillin and no pivoting: A = LU

Declaration

public void Bsrilu02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with fillin and no pivoting: A = LU

Declaration

public void Bsrilu02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02Analysis(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with 0 fillin and no pivoting: A = LU

Declaration

public void Bsrilu02Analysis(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrilu02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Bsrilu02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrilu02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Bsrilu02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrilu02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Bsrilu02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrilu02BufferSize(cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Bsrilu02BufferSize(cusparseDirection dirA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrilu02Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnz (= bsrRowPtrA(m)-bsrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnz (= bsrRowPtrA(m) - bsrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of bsrColIndA gives the number nzz passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrilu02NumericBoost(bsrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<cuDoubleComplex> boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<cuDoubleComplex>	boost_val	boost value to replace a numerical zero.

Bsrilu02NumericBoost(bsrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuFloatComplex>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<cuFloatComplex> boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<cuFloatComplex>	boost_val	boost value to replace a numerical zero.

Bsrilu02NumericBoost(bsrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<double> boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<System.Double>	boost_val	boost value to replace a numerical zero.

Bsrilu02NumericBoost(bsrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Single>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<float> boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<System.Single>	boost_val	boost value to replace a numerical zero.

Bsrilu02NumericBoost(bsrilu02Info, Int32, ref Double, ref cuDoubleComplex)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, ref double tol, ref cuDoubleComplex boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
cuDoubleComplex	boost_val	boost value to replace a numerical zero.

Bsrilu02NumericBoost(bsrilu02Info, Int32, ref Double, ref cuFloatComplex)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, ref double tol, ref cuFloatComplex boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
cuFloatComplex	boost_val	boost value to replace a numerical zero.

Bsrilu02NumericBoost(bsrilu02Info, Int32, ref Double, ref Double)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, ref double tol, ref double boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
System.Double	boost_val	boost value to replace a numerical zero.

Bsrilu02NumericBoost(bsrilu02Info, Int32, ref Double, ref Single)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling bsrilu02(). To disable a boost value, the user can call bsrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Bsrilu02NumericBoost(bsrilu02Info info, int enable_boost, ref double tol, ref float boost_val)

Parameters

Type	Name	Description
bsrilu02Info	info	structure initialized using cusparseCreateBsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
System.Single	boost_val	boost value to replace a numerical zero.

Bsrilu02ZeroPivot(CudaSparseBsrilu02Info, CudaDeviceVariable<Int32>)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXbsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Bsrilu02ZeroPivot(CudaSparseBsrilu02Info info, CudaDeviceVariable<int> position)

Parameters

Type	Name	Description
CudaSparseBsrilu02Info	info	info contains structural zero or numerical zero if the user already called bsrsv2_analysis() or bsrsv2_solve().
CudaDeviceVariable<System.Int32>	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Bsrilu02ZeroPivot(CudaSparseBsrilu02Info, ref Int32)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXbsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Bsrilu02ZeroPivot(CudaSparseBsrilu02Info info, ref int position)

Parameters

Type	Name	Description
CudaSparseBsrilu02Info	info	info contains structural zero or numerical zero if the user already called bsrsv2_analysis() or bsrsv2_solve().
System.Int32	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Double>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Double>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Single>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Single>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<cuDoubleComplex> B, int ldb, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
cuDoubleComplex	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<cuFloatComplex> B, int ldb, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
cuFloatComplex	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<double> B, int ldb, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Double>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
System.Double	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Double>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmm(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs one of the following matrix-matrix operations: C = alpha * op(A) * op(B) + beta * C

Declaration

public void Bsrmm(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transB, int mb, int n, int kb, int nnzb, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockSize, CudaDeviceVariable<float> B, int ldb, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	n	number of columns of dense matrix op(B) and A.
System.Int32	kb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of non-zero blocks of sparse matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb + 1elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Single>	B	array of dimensions (ldb, n) if op(B)=B and (ldb, k) otherwise.
System.Int32	ldb	leading dimension of B. If op(B)=B, it must be at least max(l,k) If op(B) != B, it must be at least max(1, n).
System.Single	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Single>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max(l,m) if op(A)=A and at least max(l,k) otherwise.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuDoubleComplex>	x	vector of nb*blockDim elements.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of mb*blockDim element.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuFloatComplex>	x	vector of nb*blockDim elements.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of mb*blockDim element.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<double> x, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Double>	x	vector of nb*blockDim elements.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of mb*blockDim element.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<float> x, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Single>	x	vector of nb*blockDim elements.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of mb*blockDim element.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuDoubleComplex>, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<cuDoubleComplex> x, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuDoubleComplex>	x	vector of nb*blockDim elements.
cuDoubleComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of mb*blockDim element.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<cuFloatComplex>, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<cuFloatComplex> x, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<cuFloatComplex>	x	vector of nb*blockDim elements.
cuFloatComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of mb*blockDim element.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Double>, Double, CudaDeviceVariable<Double>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<double> x, double beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Double>	x	vector of nb*blockDim elements.
System.Double	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of mb*blockDim element.

Bsrmv(cusparseDirection, cusparseOperation, Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Single>, Single, CudaDeviceVariable<Single>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + beta * y

where A is (mbblockDim) x (nbblockDim) sparse matrix (that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd), x and y are vectors, alpha and beta are scalars.

Declaration

public void Bsrmv(cusparseDirection dirA, cusparseOperation transA, int mb, int nb, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaDeviceVariable<float> x, float beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A). Only CUSPARSE_OPERATION_NON_TRANSPOSE is supported.
System.Int32	mb	number of block rows of matrix A.
System.Int32	nb	number of block columns of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb (= bsrRowPtr(mb) - bsrRowPtr(0)) non-zero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtr(m) - bsrRowPtr(0)) column indices of the non-zero blocks of matrix A. Length of bsrColIndA gives the number nzzb passed to CUSPARSE.
System.Int32	blockDim	block dimension of sparse matrix A, larger than zero.
CudaDeviceVariable<System.Single>	x	vector of nb*blockDim elements.
System.Single	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of mb*blockDim element.

Bsrsm2Analysis(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsm2(), a new sparse triangular linear system op(A)*op(Y) = alpha op(X).

Declaration

public void Bsrsm2Analysis(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is return by bsrsm2_bufferSizeExt().

Bsrsm2Analysis(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsm2(), a new sparse triangular linear system op(A)*op(Y) = alpha op(X).

Declaration

public void Bsrsm2Analysis(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is return by bsrsm2_bufferSizeExt().

Bsrsm2Analysis(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsm2(), a new sparse triangular linear system op(A)*op(Y) = alpha op(X).

Declaration

public void Bsrsm2Analysis(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is return by bsrsm2_bufferSizeExt().

Bsrsm2Analysis(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsm2(), a new sparse triangular linear system op(A)*op(Y) = alpha op(X).

Declaration

public void Bsrsm2Analysis(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is return by bsrsm2_bufferSizeExt().

Bsrsm2BufferSize(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info)

This function returns size of buffer used in bsrsm2(), a new sparse triangular linear system op(A)*Y = alpha op(X).

Declaration

public SizeT Bsrsm2BufferSize(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.

Returns

Type	Description
SizeT	number of bytes of the buffer used in bsrsm2_analysis() and bsrsm2_solve().

Bsrsm2BufferSize(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info)

This function returns size of buffer used in bsrsm2(), a new sparse triangular linear system op(A)*Y = alpha op(X).

Declaration

public SizeT Bsrsm2BufferSize(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.

Returns

Type	Description
SizeT	number of bytes of the buffer used in bsrsm2_analysis() and bsrsm2_solve().

Bsrsm2BufferSize(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info)

This function returns size of buffer used in bsrsm2(), a new sparse triangular linear system op(A)*Y = alpha op(X).

Declaration

public SizeT Bsrsm2BufferSize(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.

Returns

Type	Description
SizeT	number of bytes of the buffer used in bsrsm2_analysis() and bsrsm2_solve().

Bsrsm2BufferSize(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info)

This function returns size of buffer used in bsrsm2(), a new sparse triangular linear system op(A)*Y = alpha op(X).

Declaration

public SizeT Bsrsm2BufferSize(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(X).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of matrix Y and op(X).
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, while the supported diagonal types are CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrVal	array of nnzb bsrRowPtrA(mb) bsrRowPtrA(0) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb +1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A; larger than zero.
CudaSparseBsrsm2Info	info	record internal states based on different algorithms.

Returns

Type	Description
SizeT	number of bytes of the buffer used in bsrsm2_analysis() and bsrsm2_solve().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<cuDoubleComplex> X, int ldx, CudaDeviceVariable<cuDoubleComplex> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<cuDoubleComplex>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<cuDoubleComplex>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<cuFloatComplex> X, int ldx, CudaDeviceVariable<cuFloatComplex> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<cuFloatComplex>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<cuFloatComplex>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<double> X, int ldx, CudaDeviceVariable<double> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<System.Double>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<System.Double>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<float> X, int ldx, CudaDeviceVariable<float> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<System.Single>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<System.Single>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<cuDoubleComplex> X, int ldx, CudaDeviceVariable<cuDoubleComplex> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<cuDoubleComplex>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<cuDoubleComplex>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<cuFloatComplex> X, int ldx, CudaDeviceVariable<cuFloatComplex> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<cuFloatComplex>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<cuFloatComplex>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<double> X, int ldx, CudaDeviceVariable<double> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<System.Double>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<System.Double>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsm2Solve(cusparseDirection, cusparseOperation, cusparseOperation, Int32, Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsm2Info, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the solution of a sparse triangular linear system: op(A) * op(Y) = alpha * op(X)

Declaration

public void Bsrsm2Solve(cusparseDirection dirA, cusparseOperation transA, cusparseOperation transXY, int mb, int n, int nnzb, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int blockSize, CudaSparseBsrsm2Info info, CudaDeviceVariable<float> X, int ldx, CudaDeviceVariable<float> Y, int ldy, cusparseSolvePolicy policy, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
cusparseOperation	transA	the operation op(A).
cusparseOperation	transXY	the operation op(x) and op(Y).
System.Int32	mb	number of block rows of matrix A.
System.Int32	n	number of columns of dense matrix Y and op(X).
System.Int32	nnzb	number of non-zero blocks of matrix A
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrVal	array of nnzb ( = bsrRowPtrA(mb) - bsrRowPtrA(0) ) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb ( =bsrRowPtrA(mb) - bsrRowPtrA(0) ) column indices of the nonzero blocks of matrix A.
System.Int32	blockSize	block dimension of sparse matrix A, larger than zero.
CudaSparseBsrsm2Info	info	structure initialized using cusparseCreateBsrsm2Info().
CudaDeviceVariable<System.Single>	X	right-hand-side array.
System.Int32	ldx	leading dimension of X. If op(X)=X, ldx>=k; otherwise, ldx>=n.
CudaDeviceVariable<System.Single>	Y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y. If op(A)=A, then ldc>=m. If op(A)!=A, then ldx>=k.
cusparseSolvePolicy	policy	the supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by bsrsm2_bufferSizeExt().

Bsrsv2Analysis(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Analysis(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrsv2Analysis(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Analysis(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrsv2Analysis(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Analysis(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrsv2Analysis(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Analysis(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().

Bsrsv2BufferSize(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info)

This function returns the size of the buffer used in bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Bsrsv2BufferSize(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrsv2BufferSize(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info)

This function returns the size of the buffer used in bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Bsrsv2BufferSize(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrsv2BufferSize(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info)

This function returns the size of the buffer used in bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Bsrsv2BufferSize(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrsv2BufferSize(cusparseOperation, cusparseDirection, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info)

This function returns the size of the buffer used in bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Bsrsv2BufferSize(cusparseOperation transA, cusparseDirection dirA, int mb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<cuDoubleComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuDoubleComplex>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<cuDoubleComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<cuFloatComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuFloatComplex>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<cuFloatComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<Double>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Double>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<double> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<System.Double>	y	solution vector of size m.

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<Single>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Single>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<float> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<System.Single>	y	solution vector of size m.

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, ref cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<cuDoubleComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuDoubleComplex>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, ref cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<cuDoubleComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, ref cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<cuFloatComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuFloatComplex>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, ref cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<cuFloatComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, ref Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<Double>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Double>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, ref double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<double> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<System.Double>	y	solution vector of size m.

Bsrsv2Solve(cusparseOperation, cusparseDirection, Int32, ref Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseBsrsv2Info, CudaDeviceVariable<Single>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Single>)

This function performs the solve phase of bsrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Bsrsv2Solve(cusparseOperation transA, cusparseDirection dirA, int mb, ref float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int blockDim, CudaSparseBsrsv2Info info, CudaDeviceVariable<float> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) nonzero blocks of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of m + 1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb (= bsrRowPtrA(mb) - bsrRowPtrA(0)) column indices of the nonzero blocks of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A; must be larger than zero.
CudaSparseBsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by bsrsv2_bufferSizeExt().
CudaDeviceVariable<System.Single>	y	solution vector of size m.

Bsrsv2ZeroPivot(CudaSparseBsrsv2Info, CudaDeviceVariable<Int32>)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXbsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Bsrsv2ZeroPivot(CudaSparseBsrsv2Info info, CudaDeviceVariable<int> position)

Parameters

Type	Name	Description
CudaSparseBsrsv2Info	info	info contains structural zero or numerical zero if the user already called bsrsv2_analysis() or bsrsv2_solve().
CudaDeviceVariable<System.Int32>	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Bsrsv2ZeroPivot(CudaSparseBsrsv2Info, ref Int32)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXbsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Bsrsv2ZeroPivot(CudaSparseBsrsv2Info info, ref int position)

Parameters

Type	Name	Description
CudaSparseBsrsv2Info	info	info contains structural zero or numerical zero if the user already called bsrsv2_analysis() or bsrsv2_solve().
System.Int32	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

CoosortBufferSize(Int32, Int32, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function sorts COO format. The stable sorting is in-place. Also the user can sort by row or sort by column.

A is an m x n sparse matrix that is defined in COO storage format by the three arrays cooVals, cooRows, and cooCols.

The matrix must be base 0.

Declaration

public SizeT CoosortBufferSize(int m, int n, int nnz, CudaDeviceVariable<int> cooRowsA, CudaDeviceVariable<int> cooColsA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	cooRowsA	integer array of nnz unsorted row indices of A.
CudaDeviceVariable<System.Int32>	cooColsA	integer array of nnz unsorted column indices of A.

Returns

Type	Description
SizeT	number of bytes of the buffer.

CoosortByColumn(Int32, Int32, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

This function sorts COO format. The stable sorting is in-place. Also the user can sort by row or sort by column.

A is an m x n sparse matrix that is defined in COO storage format by the three arrays cooVals, cooRows, and cooCols.

The matrix must be base 0.

Declaration

public void CoosortByColumn(int m, int n, int nnz, CudaDeviceVariable<int> cooRowsA, CudaDeviceVariable<int> cooColsA, CudaDeviceVariable<int> P, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	cooRowsA	integer array of nnz unsorted row indices of A.
CudaDeviceVariable<System.Int32>	cooColsA	integer array of nnz unsorted column indices of A.
CudaDeviceVariable<System.Int32>	P	integer array of nnz sorted map indices.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by CoosortBufferSize().

CoosortByRow(Int32, Int32, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

This function sorts COO format. The stable sorting is in-place. Also the user can sort by row or sort by column.

A is an m x n sparse matrix that is defined in COO storage format by the three arrays cooVals, cooRows, and cooCols.

The matrix must be base 0.

Declaration

public void CoosortByRow(int m, int n, int nnz, CudaDeviceVariable<int> cooRowsA, CudaDeviceVariable<int> cooColsA, CudaDeviceVariable<int> P, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	cooRowsA	integer array of nnz unsorted row indices of A.
CudaDeviceVariable<System.Int32>	cooColsA	integer array of nnz unsorted column indices of A.
CudaDeviceVariable<System.Int32>	P	integer array of nnz sorted map indices.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by CoosortBufferSize().

CreateIdentityPermutation(Int32, CudaDeviceVariable<Int32>)

This function creates an identity map. The output parameter p represents such map by p = 0:1:(n-1).

This function is typically used with coosort, csrsort, cscsort, csr2csc_indexOnly.

Declaration

public void CreateIdentityPermutation(int n, CudaDeviceVariable<int> p)

Parameters

Type	Name	Description
System.Int32	n	size of the map.
CudaDeviceVariable<System.Int32>	p	integer array of dimensions n.

Csc2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This routine converts a sparse matrix in CSC storage format to a dense matrix.

Declaration

public void Csc2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	cscValA	array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
CudaDeviceVariable<cuDoubleComplex>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of dense array A.

Csc2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, Int32)

This routine converts a sparse matrix in CSC storage format to a dense matrix.

Declaration

public void Csc2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	cscValA	array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
CudaDeviceVariable<cuFloatComplex>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of dense array A.

Csc2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Int32)

This routine converts a sparse matrix in CSC storage format to a dense matrix.

Declaration

public void Csc2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	cscValA	array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
CudaDeviceVariable<System.Double>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of dense array A.

Csc2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Int32)

This routine converts a sparse matrix in CSC storage format to a dense matrix.

Declaration

public void Csc2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	cscValA	array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
CudaDeviceVariable<System.Single>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of dense array A.

Csc2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

Declaration

public void Csc2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<cuDoubleComplex>	cscValA
CudaDeviceVariable<System.Int32>	cscRowIndA
CudaDeviceVariable<System.Int32>	cscColPtrA
CudaSparseHybMat	hybA
System.Int32	userEllWidth
cusparseHybPartition	partitionType

Csc2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

Declaration

public void Csc2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<cuFloatComplex>	cscValA
CudaDeviceVariable<System.Int32>	cscRowIndA
CudaDeviceVariable<System.Int32>	cscColPtrA
CudaSparseHybMat	hybA
System.Int32	userEllWidth
cusparseHybPartition	partitionType

Csc2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

Declaration

public void Csc2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	cscValA
CudaDeviceVariable<System.Int32>	cscRowIndA
CudaDeviceVariable<System.Int32>	cscColPtrA
CudaSparseHybMat	hybA
System.Int32	userEllWidth
cusparseHybPartition	partitionType

Csc2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

Declaration

public void Csc2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	cscValA
CudaDeviceVariable<System.Int32>	cscRowIndA
CudaDeviceVariable<System.Int32>	cscColPtrA
CudaSparseHybMat	hybA
System.Int32	userEllWidth
cusparseHybPartition	partitionType

Cscsort(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

This function sorts CSC format. The stable sorting is in-place.

The matrix type is regarded as CUSPARSE_MATRIX_TYPE_GENERAL implicitly. In other words, any symmetric property is ignored.

This function cscsort() requires buffer size returned by cscsort_bufferSizeExt(). The address of pBuffer must be multiple of 128 bytes. If not, CUSPARSE_STATUS_INVALID_VALUE is returned.

The parameter P is both input and output. If the user wants to compute sorted cscVal, P must be set as 0:1:(nnz-1) before cscsort(), and after cscsort(), new sorted value array satisfies cscVal_sorted = cscVal(P).

Declaration

public void Cscsort(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> cscColPtrA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> P, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
CudaDeviceVariable<System.Int32>	cscRowIndA	integer array of nnz unsorted row indices of A.
CudaDeviceVariable<System.Int32>	P	integer array of nnz sorted map indices.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by CscsortBufferSize().

CscsortBufferSize(Int32, Int32, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function sorts CSC format. The stable sorting is in-place.

The matrix type is regarded as CUSPARSE_MATRIX_TYPE_GENERAL implicitly. In other words, any symmetric property is ignored.

This function cscsort() requires buffer size returned by cscsort_bufferSizeExt(). The address of pBuffer must be multiple of 128 bytes. If not, CUSPARSE_STATUS_INVALID_VALUE is returned.

The parameter P is both input and output. If the user wants to compute sorted cscVal, P must be set as 0:1:(nnz-1) before cscsort(), and after cscsort(), new sorted value array satisfies cscVal_sorted = cscVal(P).

Declaration

public SizeT CscsortBufferSize(int m, int n, int nnz, CudaDeviceVariable<int> cscColPtrA, CudaDeviceVariable<int> cscRowIndA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
CudaDeviceVariable<System.Int32>	cscRowIndA	integer array of nnz unsorted row indices of A.

Returns

Type	Description
SizeT	number of bytes of the buffer.

Csr2bsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in CSR format (that is defined by the three arrays csrValA, csrRowPtrA and csrColIndA) into a sparse matrix in BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC). A is m x n sparse matrix and C is (mbblockDim) x (nbblockDim) sparse matrix.

Declaration

public void Csr2bsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A. The range of blockDim is between 1 and min(m, n).
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<cuDoubleComplex>	bsrValC	array of nnzb*blockDim² non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzb column indices of the non-zero blocks of matrix C.

Csr2bsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in CSR format (that is defined by the three arrays csrValA, csrRowPtrA and csrColIndA) into a sparse matrix in BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC). A is m x n sparse matrix and C is (mbblockDim) x (nbblockDim) sparse matrix.

Declaration

public void Csr2bsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A. The range of blockDim is between 1 and min(m, n).
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<cuFloatComplex>	bsrValC	array of nnzb*blockDim² non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzb column indices of the non-zero blocks of matrix C.

Csr2bsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in CSR format (that is defined by the three arrays csrValA, csrRowPtrA and csrColIndA) into a sparse matrix in BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC). A is m x n sparse matrix and C is (mbblockDim) x (nbblockDim) sparse matrix.

Declaration

public void Csr2bsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A. The range of blockDim is between 1 and min(m, n).
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<System.Double>	bsrValC	array of nnzb*blockDim² non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzb column indices of the non-zero blocks of matrix C.

Csr2bsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in CSR format (that is defined by the three arrays csrValA, csrRowPtrA and csrColIndA) into a sparse matrix in BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC). A is m x n sparse matrix and C is (mbblockDim) x (nbblockDim) sparse matrix.

Declaration

public void Csr2bsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A. The range of blockDim is between 1 and min(m, n).
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<System.Single>	bsrValC	array of nnzb*blockDim² non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzb column indices of the non-zero blocks of matrix C.

Csr2bsrNnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in CSR format (that is defined by the three arrays csrValA, csrRowPtrA and csrColIndA) into a sparse matrix in BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC). A is m x n sparse matrix and C is (mbblockDim) x (nbblockDim) sparse matrix.

Declaration

public void Csr2bsrNnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A. The range of blockDim is between 1 and min(m, n).
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

Csr2bsrNnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, ref Int32)

This function converts a sparse matrix in CSR format (that is defined by the three arrays csrValA, csrRowPtrA and csrColIndA) into a sparse matrix in BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC). A is m x n sparse matrix and C is (mbblockDim) x (nbblockDim) sparse matrix.

Declaration

public void Csr2bsrNnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, int blockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> bsrRowPtrC, ref int nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
System.Int32	blockDim	block dimension of sparse matrix A. The range of blockDim is between 1 and min(m, n).
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
System.Int32	nnzTotalDevHostPtr

Csr2csc(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase)

This routine converts a matrix from CSR to CSC sparse storage format. The resulting matrix can be re-interpreted as a transpose of the original matrix in CSR storage format.

Declaration

public void Csr2csc(int m, int n, CudaDeviceVariable<cuDoubleComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaDeviceVariable<cuDoubleComplex> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr, cusparseAction copyValues, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	csrVal	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	cscVal	Output: array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) nonzero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	cscRowInd	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtr	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	idxBase	Index base.

Csr2csc(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase)

This routine converts a matrix from CSR to CSC sparse storage format. The resulting matrix can be re-interpreted as a transpose of the original matrix in CSR storage format.

Declaration

public void Csr2csc(int m, int n, CudaDeviceVariable<cuFloatComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaDeviceVariable<cuFloatComplex> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr, cusparseAction copyValues, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	csrVal	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	cscVal	Output: array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) nonzero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	cscRowInd	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtr	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	idxBase	Index base.

Csr2csc(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase)

This routine converts a matrix from CSR to CSC sparse storage format. The resulting matrix can be re-interpreted as a transpose of the original matrix in CSR storage format.

Declaration

public void Csr2csc(int m, int n, CudaDeviceVariable<double> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaDeviceVariable<double> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr, cusparseAction copyValues, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Double>	csrVal	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	cscVal	Output: array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) nonzero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	cscRowInd	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtr	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	idxBase	Index base.

Csr2csc(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase)

This routine converts a matrix from CSR to CSC sparse storage format. The resulting matrix can be re-interpreted as a transpose of the original matrix in CSR storage format.

Declaration

public void Csr2csc(int m, int n, CudaDeviceVariable<float> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaDeviceVariable<float> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr, cusparseAction copyValues, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Single>	csrVal	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	cscVal	Output: array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) nonzero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	cscRowInd	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtr	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	idxBase	Index base.

Csr2csr_compress(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuDoubleComplex)

This routine takes as input a csr form where the values may have 0 elements and compresses it to return a csr form with no zeros.

Declaration

public void Csr2csr_compress(int m, int n, CudaSparseMatrixDescriptor descra, CudaDeviceVariable<cuDoubleComplex> inVal, CudaDeviceVariable<int> inColInd, CudaDeviceVariable<int> inRowPtr, int inNnz, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<cuDoubleComplex> outVal, CudaDeviceVariable<int> outColInd, CudaDeviceVariable<int> outRowPtr, cuDoubleComplex tol)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descra
CudaDeviceVariable<cuDoubleComplex>	inVal
CudaDeviceVariable<System.Int32>	inColInd
CudaDeviceVariable<System.Int32>	inRowPtr
System.Int32	inNnz
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<cuDoubleComplex>	outVal
CudaDeviceVariable<System.Int32>	outColInd
CudaDeviceVariable<System.Int32>	outRowPtr
cuDoubleComplex	tol

Csr2csr_compress(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuFloatComplex)

This routine takes as input a csr form where the values may have 0 elements and compresses it to return a csr form with no zeros.

Declaration

public void Csr2csr_compress(int m, int n, CudaSparseMatrixDescriptor descra, CudaDeviceVariable<cuFloatComplex> inVal, CudaDeviceVariable<int> inColInd, CudaDeviceVariable<int> inRowPtr, int inNnz, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<cuFloatComplex> outVal, CudaDeviceVariable<int> outColInd, CudaDeviceVariable<int> outRowPtr, cuFloatComplex tol)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descra
CudaDeviceVariable<cuFloatComplex>	inVal
CudaDeviceVariable<System.Int32>	inColInd
CudaDeviceVariable<System.Int32>	inRowPtr
System.Int32	inNnz
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<cuFloatComplex>	outVal
CudaDeviceVariable<System.Int32>	outColInd
CudaDeviceVariable<System.Int32>	outRowPtr
cuFloatComplex	tol

Csr2csr_compress(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double)

This routine takes as input a csr form where the values may have 0 elements and compresses it to return a csr form with no zeros.

Declaration

public void Csr2csr_compress(int m, int n, CudaSparseMatrixDescriptor descra, CudaDeviceVariable<double> inVal, CudaDeviceVariable<int> inColInd, CudaDeviceVariable<int> inRowPtr, int inNnz, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<double> outVal, CudaDeviceVariable<int> outColInd, CudaDeviceVariable<int> outRowPtr, double tol)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descra
CudaDeviceVariable<System.Double>	inVal
CudaDeviceVariable<System.Int32>	inColInd
CudaDeviceVariable<System.Int32>	inRowPtr
System.Int32	inNnz
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<System.Double>	outVal
CudaDeviceVariable<System.Int32>	outColInd
CudaDeviceVariable<System.Int32>	outRowPtr
System.Double	tol

Csr2csr_compress(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single)

This routine takes as input a csr form where the values may have 0 elements and compresses it to return a csr form with no zeros.

Declaration

public void Csr2csr_compress(int m, int n, CudaSparseMatrixDescriptor descra, CudaDeviceVariable<float> inVal, CudaDeviceVariable<int> inColInd, CudaDeviceVariable<int> inRowPtr, int inNnz, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<float> outVal, CudaDeviceVariable<int> outColInd, CudaDeviceVariable<int> outRowPtr, float tol)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaSparseMatrixDescriptor	descra
CudaDeviceVariable<System.Single>	inVal
CudaDeviceVariable<System.Int32>	inColInd
CudaDeviceVariable<System.Int32>	inRowPtr
System.Int32	inNnz
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<System.Single>	outVal
CudaDeviceVariable<System.Int32>	outColInd
CudaDeviceVariable<System.Int32>	outRowPtr
System.Single	tol

Csr2csru(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csr2csru(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csr2csru(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csr2csru(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csr2csru(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csr2csru(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csr2csru(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csr2csru(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csr2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This routine converts a sparse matrix in CSR storage format to a dense matrix.

Declaration

public void Csr2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<cuDoubleComplex>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of array matrix A.

Csr2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, Int32)

This routine converts a sparse matrix in CSR storage format to a dense matrix.

Declaration

public void Csr2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<cuFloatComplex>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of array matrix A.

Csr2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Int32)

This routine converts a sparse matrix in CSR storage format to a dense matrix.

Declaration

public void Csr2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Double>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of array matrix A.

Csr2dense(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Int32)

This routine converts a sparse matrix in CSR storage format to a dense matrix.

Declaration

public void Csr2dense(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Single>	A	Output: array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	leading dimension of array matrix A.

Csr2gebsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix A in CSR format (that is defined by arrays csrValA, csrRowPtrA, and csrColIndA) into a sparse matrix C in general BSR format (that is defined by the three arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Csr2gebsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDim, int colBlockDim, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValC	array of nnzbrowBlockDimcolBlockDim nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	Pointer nnzTotalDevHostPtr can point to a device memory or host memory.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by csr2gebsr_bufferSizeExt().

Csr2gebsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix A in CSR format (that is defined by arrays csrValA, csrRowPtrA, and csrColIndA) into a sparse matrix C in general BSR format (that is defined by the three arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Csr2gebsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDim, int colBlockDim, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValC	array of nnzbrowBlockDimcolBlockDim nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	Pointer nnzTotalDevHostPtr can point to a device memory or host memory.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by csr2gebsr_bufferSizeExt().

Csr2gebsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix A in CSR format (that is defined by arrays csrValA, csrRowPtrA, and csrColIndA) into a sparse matrix C in general BSR format (that is defined by the three arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Csr2gebsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDim, int colBlockDim, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValC	array of nnzbrowBlockDimcolBlockDim nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	Pointer nnzTotalDevHostPtr can point to a device memory or host memory.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by csr2gebsr_bufferSizeExt().

Csr2gebsr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix A in CSR format (that is defined by arrays csrValA, csrRowPtrA, and csrColIndA) into a sparse matrix C in general BSR format (that is defined by the three arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Csr2gebsr(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDim, int colBlockDim, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValC	array of nnzbrowBlockDimcolBlockDim nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	Pointer nnzTotalDevHostPtr can point to a device memory or host memory.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by csr2gebsr_bufferSizeExt().

Csr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns the size of the buffer used in computing csr2gebsrNnz and csr2gebsr.

Declaration

public SizeT Csr2gebsrBufferSize(cusparseDirection dir, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
cusparseDirection	dir	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrVal	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz column indices of the nonzero elements of matrix A.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Csr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns the size of the buffer used in computing csr2gebsrNnz and csr2gebsr.

Declaration

public SizeT Csr2gebsrBufferSize(cusparseDirection dir, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
cusparseDirection	dir	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrVal	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz column indices of the nonzero elements of matrix A.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Csr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns the size of the buffer used in computing csr2gebsrNnz and csr2gebsr.

Declaration

public SizeT Csr2gebsrBufferSize(cusparseDirection dir, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
cusparseDirection	dir	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrVal	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz column indices of the nonzero elements of matrix A.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Csr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns the size of the buffer used in computing csr2gebsrNnz and csr2gebsr.

Declaration

public SizeT Csr2gebsrBufferSize(cusparseDirection dir, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
cusparseDirection	dir	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrVal	array of nnz nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz column indices of the nonzero elements of matrix A.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Csr2gebsrNnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

This function converts a sparse matrix A in CSR format (that is defined by arrays csrValA, csrRowPtrA, and csrColIndA) into a sparse matrix C in general BSR format (that is defined by the three arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Csr2gebsrNnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> bsrRowPtrC, int rowBlockDim, int colBlockDim, CudaDeviceVariable<int> nnzTotalDevHostPtr, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr	total number of nonzero blocks of matrix C. Pointer nnzTotalDevHostPtr can point to a device memory or host memory.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by csr2gebsr_bufferSizeExt().

Csr2gebsrNnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, Int32, Int32, ref Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix A in CSR format (that is defined by arrays csrValA, csrRowPtrA, and csrColIndA) into a sparse matrix C in general BSR format (that is defined by the three arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Csr2gebsrNnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> bsrRowPtrC, int rowBlockDim, int colBlockDim, ref int nnzTotalDevHostPtr, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix A
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
System.Int32	rowBlockDim	number of rows within a block of C.
System.Int32	colBlockDim	number of columns within a block of C.
System.Int32	nnzTotalDevHostPtr	total number of nonzero blocks of matrix C. Pointer nnzTotalDevHostPtr can point to a device memory or host memory.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by csr2gebsr_bufferSizeExt().

Csr2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a sparse matrix in CSR storage format to a sparse matrix in HYB storage format.

Declaration

public void Csr2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A in CSR format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Csr2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a sparse matrix in CSR storage format to a sparse matrix in HYB storage format.

Declaration

public void Csr2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A in CSR format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Csr2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a sparse matrix in CSR storage format to a sparse matrix in HYB storage format.

Declaration

public void Csr2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A in CSR format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Csr2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a sparse matrix in CSR storage format to a sparse matrix in HYB storage format.

Declaration

public void Csr2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A in CSR format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaDeviceVariable<double> fractionToColor, CudaDeviceVariable<int> ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
CudaDeviceVariable<System.Double>	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
CudaDeviceVariable<System.Int32>	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, ref Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, double fractionToColor, ref int ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
System.Double	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
System.Int32	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaDeviceVariable<float> fractionToColor, CudaDeviceVariable<int> ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
CudaDeviceVariable<System.Single>	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
CudaDeviceVariable<System.Int32>	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, ref Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, float fractionToColor, ref int ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
System.Single	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
System.Int32	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaDeviceVariable<double> fractionToColor, CudaDeviceVariable<int> ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
CudaDeviceVariable<System.Double>	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
CudaDeviceVariable<System.Int32>	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, ref Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, double fractionToColor, ref int ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
System.Double	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
System.Int32	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaDeviceVariable<float> fractionToColor, CudaDeviceVariable<int> ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
CudaDeviceVariable<System.Single>	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
CudaDeviceVariable<System.Int32>	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrcolor(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, ref Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseColorInfo)

This function performs the coloring of the adjacency graph associated with the matrix A stored in CSR format. The coloring is an assignment of colors (integer numbers) to nodes, such that neighboring nodes have distinct colors. An approximate coloring algorithm is used in this routine, and is stopped when a certain percentage of nodes has been colored. The rest of the nodes are assigned distinct colors (an increasing sequence of integers numbers, starting from the last integer used previously). The last two auxiliary routines can be used to extract the resulting number of colors, their assignment and the associated reordering. The reordering is such that nodes that have been assigned the same color are reordered to be next to each other.

The matrix A passed to this routine, must be stored as a general matrix and have a symmetric sparsity pattern. If the matrix is nonsymmetric the user should pass A+A^T as a parameter to this routine.

Declaration

public void Csrcolor(int m, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, float fractionToColor, ref int ncolors, CudaDeviceVariable<int> coloring, CudaDeviceVariable<int> reordering, CudaSparseColorInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrSortedValA	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix A.
System.Single	fractionToColor	fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.
System.Int32	ncolors	The number of distinct colors used (at most the size of the matrix, but likely much smaller).
CudaDeviceVariable<System.Int32>	coloring	The resulting coloring permutation.
CudaDeviceVariable<System.Int32>	reordering	The resulting reordering permutation (untouched if NULL)
CudaSparseColorInfo	info	structure with information to be passed to the coloring.

Csrgeam(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<cuDoubleComplex> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgeam(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<cuFloatComplex> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgeam(Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<double> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgeam(Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<float> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgeam(Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, cuDoubleComplex beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<cuDoubleComplex> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
cuDoubleComplex	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgeam(Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, cuFloatComplex beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<cuFloatComplex> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
cuFloatComplex	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgeam(Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, double beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<double> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
System.Double	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgeam(Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void Csrgeam(int m, int n, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float beta, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<float> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
System.Single	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

CsrgeamNnz(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void CsrgeamNnz(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

CsrgeamNnz(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, ref Int32)

This function performs following matrix-matrix operation

C = alpha * A + beta * B

where A, B and C are m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC respectively), and alpha and beta are scalars. Since A and B have different sparsity patterns, CUSPARSE adopts two-step approach to complete sparse matrix C.

Declaration

public void CsrgeamNnz(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, ref int nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A,B,C.
System.Int32	n	number of columns of sparse matrix A,B,C.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
System.Int32	nnzTotalDevHostPtr

Csrgemm(cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = op(A) * op(B)

where op(A), op(B) and C are m x k, k x n, and m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC) respectively.

Only support devices of compute capability 2.0 or above.

Declaration

public void Csrgemm(cusparseOperation transA, cusparseOperation transB, int m, int n, int k, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<cuDoubleComplex> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	m	number of rows of sparse matrix op(A) and C.
System.Int32	n	number of columns of sparse matrix op(B) and C.
System.Int32	k	number of columns/rows of sparse matrix op(A) / op(B).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of ~m + 1 elements that contains the start of every row and the end of the last row plus one. ~m = m if transA == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~m = k.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of ~k + 1 elements that contains the start of every row and the end of the last row plus one. ~k = k if transB == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~k = n.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgemm(cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = op(A) * op(B)

where op(A), op(B) and C are m x k, k x n, and m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC) respectively.

Only support devices of compute capability 2.0 or above.

Declaration

public void Csrgemm(cusparseOperation transA, cusparseOperation transB, int m, int n, int k, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<cuFloatComplex> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	m	number of rows of sparse matrix op(A) and C.
System.Int32	n	number of columns of sparse matrix op(B) and C.
System.Int32	k	number of columns/rows of sparse matrix op(A) / op(B).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of ~m + 1 elements that contains the start of every row and the end of the last row plus one. ~m = m if transA == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~m = k.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of ~k + 1 elements that contains the start of every row and the end of the last row plus one. ~k = k if transB == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~k = n.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgemm(cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = op(A) * op(B)

where op(A), op(B) and C are m x k, k x n, and m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC) respectively.

Only support devices of compute capability 2.0 or above.

Declaration

public void Csrgemm(cusparseOperation transA, cusparseOperation transB, int m, int n, int k, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<double> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	m	number of rows of sparse matrix op(A) and C.
System.Int32	n	number of columns of sparse matrix op(B) and C.
System.Int32	k	number of columns/rows of sparse matrix op(A) / op(B).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of ~m + 1 elements that contains the start of every row and the end of the last row plus one. ~m = m if transA == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~m = k.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of ~k + 1 elements that contains the start of every row and the end of the last row plus one. ~k = k if transB == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~k = n.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgemm(cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = op(A) * op(B)

where op(A), op(B) and C are m x k, k x n, and m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC) respectively.

Only support devices of compute capability 2.0 or above.

Declaration

public void Csrgemm(cusparseOperation transA, cusparseOperation transB, int m, int n, int k, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<float> csrValB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	m	number of rows of sparse matrix op(A) and C.
System.Int32	n	number of columns of sparse matrix op(B) and C.
System.Int32	k	number of columns/rows of sparse matrix op(A) / op(B).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValA	array of nnzA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of ~m + 1 elements that contains the start of every row and the end of the last row plus one. ~m = m if transA == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~m = k.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValB	array of nnzB non-zero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of ~k + 1 elements that contains the start of every row and the end of the last row plus one. ~k = k if transB == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~k = n.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrValC	array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnzC (= csrRowPtrC(m) - csrRowPtrC(0)) column indices of the non-zero elements of matrix C.

Csrgemm2(Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<cuDoubleComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<cuDoubleComplex> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<cuDoubleComplex> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<cuDoubleComplex> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2(Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<cuFloatComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<cuFloatComplex> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<cuFloatComplex> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<cuFloatComplex> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<cuFloatComplex>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<cuFloatComplex>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<cuFloatComplex>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2(Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<double> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<double> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<double> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<double> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Double>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Double>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Double>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2(Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<float> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<float> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<float> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<float> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Single>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Single>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Single>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2(Int32, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuDoubleComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<cuDoubleComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<cuDoubleComplex> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, cuDoubleComplex beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<cuDoubleComplex> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
cuDoubleComplex	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuDoubleComplex>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2(Int32, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuFloatComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<cuFloatComplex> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<cuFloatComplex> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, cuFloatComplex beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<cuFloatComplex> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<cuFloatComplex>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<cuFloatComplex>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
cuFloatComplex	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<cuFloatComplex>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<cuFloatComplex>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2(Int32, Int32, Int32, Double, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, double alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<double> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<double> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, double beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<double> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Double>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Double>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
System.Double	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Double>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Double>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2(Int32, Int32, Int32, Single, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2(int m, int n, int k, float alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<float> csrSortedValA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<float> csrSortedValB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, float beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<float> csrSortedValD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrSortedValC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> csrSortedColIndC, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Single>	csrSortedValA	array of nnzA nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Single>	csrSortedValB	array of nnzB nonzero elements of matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
System.Single	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Single>	csrSortedValD	array of nnzD nonzero elements of matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Single>	csrSortedValC	array of nnzC nonzero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndC	integer array of nnzC column indices of the nonzero elements of matrix C.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2BufferSize(Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<cuDoubleComplex> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2BufferSize(Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<cuFloatComplex> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2BufferSize(Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<double> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2BufferSize(Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaDeviceVariable<float> beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2BufferSize(Int32, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuDoubleComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, cuDoubleComplex beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
cuDoubleComplex	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2BufferSize(Int32, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuFloatComplex, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, cuFloatComplex beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
cuFloatComplex	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2BufferSize(Int32, Int32, Int32, Double, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, double alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, double beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
System.Double	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2BufferSize(Int32, Int32, Int32, Single, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public SizeT Csrgemm2BufferSize(int m, int n, int k, float alpha, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, float beta, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseCsrgemm2Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
System.Single	beta	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csrgemm2Nnnz and csrgemm2.

Csrgemm2Nnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2Nnz(int m, int n, int k, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrSortedRowPtrC, CudaDeviceVariable<int> nnzTotalDevHostPtr, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr	total number of nonzero elements in device or host memory. It is equal to (csrRowPtrC(m)-csrRowPtrC(0)).
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

Csrgemm2Nnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, ref Int32, CudaSparseCsrgemm2Info, CudaDeviceVariable<Byte>)

This function performs following matrix-matrix operation:

C = alpha * A *A B + beta * D

where A, B, D and C are m×k, k×n, m×n and m×n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValD|csrValC, csrRowPtrA| csrRowPtrB|csrRowPtrD|csrRowPtrC, and csrColIndA|csrColIndB|csrColIndD|csrcolIndC respectively.

Declaration

public void Csrgemm2Nnz(int m, int n, int k, CudaSparseMatrixDescriptor descrA, int nnzA, CudaDeviceVariable<int> csrSortedRowPtrA, CudaDeviceVariable<int> csrSortedColIndA, CudaSparseMatrixDescriptor descrB, int nnzB, CudaDeviceVariable<int> csrSortedRowPtrB, CudaDeviceVariable<int> csrSortedColIndB, CudaSparseMatrixDescriptor descrD, int nnzD, CudaDeviceVariable<int> csrSortedRowPtrD, CudaDeviceVariable<int> csrSortedColIndD, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrSortedRowPtrC, ref int nnzTotalDevHostPtr, CudaSparseCsrgemm2Info info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of sparse matrix A, D and C.
System.Int32	n	number of columns of sparse matrix B, D and C.
System.Int32	k	number of columns/rows of sparse matrix A / B.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzA	number of nonzero elements of sparse matrix A.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndA	integer array of nnzA column indices of the nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only
System.Int32	nnzB	number of nonzero elements of sparse matrix B.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrB	integer array of k+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndB	integer array of nnzB column indices of the nonzero elements of matrix B.
CudaSparseMatrixDescriptor	descrD	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
System.Int32	nnzD	number of nonzero elements of sparse matrix D.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrD	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrSortedColIndD	integer array of nnzD column indices of the nonzero elements of matrix D.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrSortedRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
System.Int32	nnzTotalDevHostPtr	total number of nonzero elements in device or host memory. It is equal to (csrRowPtrC(m)-csrRowPtrC(0)).
CudaSparseCsrgemm2Info	info	structure with information used in csrgemm2Nnz and csrgemm2.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by csrgemm2BufferSize

CsrgemmNnz(cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function performs following matrix-matrix operation

C = op(A) * op(B)

where op(A), op(B) and C are m x k, k x n, and m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC) respectively.

Only support devices of compute capability 2.0 or above.

Declaration

public void CsrgemmNnz(cusparseOperation transA, cusparseOperation transB, int m, int n, int k, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	m	number of rows of sparse matrix op(A) and C.
System.Int32	n	number of columns of sparse matrix op(B) and C.
System.Int32	k	number of columns/rows of sparse matrix op(A) / op(B).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of ~m + 1 elements that contains the start of every row and the end of the last row plus one. ~m = m if transA == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~m = k.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of ~k + 1 elements that contains the start of every row and the end of the last row plus one. ~k = k if transB == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~k = n.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

CsrgemmNnz(cusparseOperation, cusparseOperation, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, ref Int32)

This function performs following matrix-matrix operation

C = op(A) * op(B)

where op(A), op(B) and C are m x k, k x n, and m x n sparse matrices (defined in CSR storage format by the three arrays csrValA|csrValB|csrValC, csrRowPtrA|csrRowPtrB|csrRowPtrC, and csrColIndA|csrColIndB|csrcolIndC) respectively.

Only support devices of compute capability 2.0 or above.

Declaration

public void CsrgemmNnz(cusparseOperation transA, cusparseOperation transB, int m, int n, int k, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseMatrixDescriptor descrB, CudaDeviceVariable<int> csrRowPtrB, CudaDeviceVariable<int> csrColIndB, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, ref int nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cusparseOperation	transB	the operation op(B).
System.Int32	m	number of rows of sparse matrix op(A) and C.
System.Int32	n	number of columns of sparse matrix op(B) and C.
System.Int32	k	number of columns/rows of sparse matrix op(A) / op(B).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_ MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of ~m + 1 elements that contains the start of every row and the end of the last row plus one. ~m = m if transA == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~m = k.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnzA column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzzA passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrB	the descriptor of matrix B. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrB	integer array of ~k + 1 elements that contains the start of every row and the end of the last row plus one. ~k = k if transB == CUSPARSE_ OPERATION_NON_TRANSPOSE, otherwise ~k = n.
CudaDeviceVariable<System.Int32>	csrColIndB	integer array of nnzB column indices of the non-zero elements of matrix B. Length of csrColIndB gives the number nzzB passed to CUSPARSE.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL only.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
System.Int32	nnzTotalDevHostPtr

Csric0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-Cholesky factorization with 0 fill-in and no pivoting

op(A) ≈ R'R

where A is m*n Hermitian/symmetric positive definite sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that only a lower or upper Hermitian/symmetric part of the matrix A is actually stored. It is overwritten by the lower or upper triangular factor R' or R, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csric0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csric0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-Cholesky factorization with 0 fill-in and no pivoting

op(A) ≈ R'R

where A is m*n Hermitian/symmetric positive definite sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that only a lower or upper Hermitian/symmetric part of the matrix A is actually stored. It is overwritten by the lower or upper triangular factor R' or R, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csric0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csric0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-Cholesky factorization with 0 fill-in and no pivoting

op(A) ≈ R'R

where A is m*n Hermitian/symmetric positive definite sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that only a lower or upper Hermitian/symmetric part of the matrix A is actually stored. It is overwritten by the lower or upper triangular factor R' or R, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csric0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csric0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-Cholesky factorization with 0 fill-in and no pivoting

op(A) ≈ R'R

where A is m*n Hermitian/symmetric positive definite sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that only a lower or upper Hermitian/symmetric part of the matrix A is actually stored. It is overwritten by the lower or upper triangular factor R' or R, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csric0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csric02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public void Csric02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csric02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Csric02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csric02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Csric02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csric02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Csric02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csric02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsric02Info)

This function returns size of buffer used in computing the incomplete-Cholesky factorization with fill-in and no pivoting: A = LL^H

Declaration

public SizeT Csric02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsric02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsric02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csric02ZeroPivot(CudaSparseCsric02Info, CudaDeviceVariable<Int32>)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXcsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Csric02ZeroPivot(CudaSparseCsric02Info info, CudaDeviceVariable<int> position)

Parameters

Type	Name	Description
CudaSparseCsric02Info	info	info contains structural zero or numerical zero if the user already called csrsv2_analysis() or csrsv2_solve().
CudaDeviceVariable<System.Int32>	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Csric02ZeroPivot(CudaSparseCsric02Info, ref Int32)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXcsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Csric02ZeroPivot(CudaSparseCsric02Info info, ref int position)

Parameters

Type	Name	Description
CudaSparseCsric02Info	info	info contains structural zero or numerical zero if the user already called csrsv2_analysis() or csrsv2_solve().
System.Int32	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Csrilu0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-LU factorization with 0 fill-in and no pivoting

op(A) ≈ LU

where A is m*n sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that the diagonal of lower triangular factor L is unitary and need not be stored. Therefore the input matrix is ovewritten with the resulting lower and upper triangular factor L and U, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csrilu0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csrilu0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-LU factorization with 0 fill-in and no pivoting

op(A) ≈ LU

where A is m*n sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that the diagonal of lower triangular factor L is unitary and need not be stored. Therefore the input matrix is ovewritten with the resulting lower and upper triangular factor L and U, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csrilu0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csrilu0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-LU factorization with 0 fill-in and no pivoting

op(A) ≈ LU

where A is m*n sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that the diagonal of lower triangular factor L is unitary and need not be stored. Therefore the input matrix is ovewritten with the resulting lower and upper triangular factor L and U, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csrilu0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csrilu0(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

This function computes the incomplete-LU factorization with 0 fill-in and no pivoting

op(A) ≈ LU

where A is m*n sparse matrix (that is defined in CSR storage format by the three arrays csrValM, csrRowPtrA and csrColIndA).

Notice that the diagonal of lower triangular factor L is unitary and need not be stored. Therefore the input matrix is ovewritten with the resulting lower and upper triangular factor L and U, respectively.

A call to this routine must be preceeded by a call to the csrsv_analysis routine. This function requires some extra storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Csrilu0(cusparseOperation trans, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	trans	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA_ValM	array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).

Csrilu02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Csrilu02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Csrilu02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Csrilu02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the solve phase of the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public void Csrilu02(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA_ValM, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA_ValM	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A. Output: matrix containing the incomplete-LU lower and upper triangular factors.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with fillin and no pivoting: A = LU

Declaration

public void Csrilu02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with fillin and no pivoting: A = LU

Declaration

public void Csrilu02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with fillin and no pivoting: A = LU

Declaration

public void Csrilu02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02Analysis(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of the incomplete-LU factorization with fillin and no pivoting: A = LU

Declaration

public void Csrilu02Analysis(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrilu02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Csrilu02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrilu02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Csrilu02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrilu02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Csrilu02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrilu02BufferSize(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrilu02Info)

This function returns size of the buffer used in computing the incomplete-LU factorization with fill-in and no pivoting: A = LU

Declaration

public SizeT Csrilu02BufferSize(int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrilu02Info info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrilu02Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrilu02NumericBoost(csrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<cuDoubleComplex> boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<cuDoubleComplex>	boost_val	boost value to replace a numerical zero.

Csrilu02NumericBoost(csrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuFloatComplex>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<cuFloatComplex> boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<cuFloatComplex>	boost_val	boost value to replace a numerical zero.

Csrilu02NumericBoost(csrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<double> boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<System.Double>	boost_val	boost value to replace a numerical zero.

Csrilu02NumericBoost(csrilu02Info, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Single>)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, CudaDeviceVariable<double> tol, CudaDeviceVariable<float> boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
CudaDeviceVariable<System.Double>	tol	tolerance to determine a numerical zero.
CudaDeviceVariable<System.Single>	boost_val	boost value to replace a numerical zero.

Csrilu02NumericBoost(csrilu02Info, Int32, ref Double, ref cuDoubleComplex)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, ref double tol, ref cuDoubleComplex boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
cuDoubleComplex	boost_val	boost value to replace a numerical zero.

Csrilu02NumericBoost(csrilu02Info, Int32, ref Double, ref cuFloatComplex)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, ref double tol, ref cuFloatComplex boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
cuFloatComplex	boost_val	boost value to replace a numerical zero.

Csrilu02NumericBoost(csrilu02Info, Int32, ref Double, ref Double)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, ref double tol, ref double boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
System.Double	boost_val	boost value to replace a numerical zero.

Csrilu02NumericBoost(csrilu02Info, Int32, ref Double, ref Single)

The user can use a boost value to replace a numerical value in incomplete LU factorization. The tol is used to determine a numerical zero, and the boost_val is used to replace a numerical zero. The behavior is

if tol >= fabs(A(j,j)), then A(j,j)=boost_val.

To enable a boost value, the user has to set parameter enable_boost to 1 before calling csrilu02(). To disable a boost value, the user can call csrilu02_numericBoost() again with parameter enable_boost=0.

If enable_boost=0, tol and boost_val are ignored.

Declaration

public void Csrilu02NumericBoost(csrilu02Info info, int enable_boost, ref double tol, ref float boost_val)

Parameters

Type	Name	Description
csrilu02Info	info	structure initialized using cusparseCreateCsrilu02Info().
System.Int32	enable_boost	disable boost by enable_boost=0; otherwise, boost is enabled.
System.Double	tol	tolerance to determine a numerical zero.
System.Single	boost_val	boost value to replace a numerical zero.

Csrilu02ZeroPivot(CudaSparseCsrilu02Info, CudaDeviceVariable<Int32>)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXcsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Csrilu02ZeroPivot(CudaSparseCsrilu02Info info, CudaDeviceVariable<int> position)

Parameters

Type	Name	Description
CudaSparseCsrilu02Info	info	info contains structural zero or numerical zero if the user already called csrsv2_analysis() or csrsv2_solve().
CudaDeviceVariable<System.Int32>	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Csrilu02ZeroPivot(CudaSparseCsrilu02Info, ref Int32)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXcsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Csrilu02ZeroPivot(CudaSparseCsrilu02Info info, ref int position)

Parameters

Type	Name	Description
CudaSparseCsrilu02Info	info	info contains structural zero or numerical zero if the user already called csrsv2_analysis() or csrsv2_solve().
System.Int32	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Csrmm(cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm(cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm(cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Double>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm(cusparseOperation, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Single>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm(cusparseOperation, Int32, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> B, int ldb, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
cuDoubleComplex	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm(cusparseOperation, Int32, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> B, int ldb, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
cuFloatComplex	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm(cusparseOperation, Int32, Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> B, int ldb, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
System.Double	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Double>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm(cusparseOperation, Int32, Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

Matrix-matrix multiplication C = alpha * op(A) * B + beta * C, where A is a sparse matrix, B and C are dense and usually tall matrices.

Declaration

public void Csrmm(cusparseOperation transA, int m, int n, int k, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> B, int ldb, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of sparse matrix A.
System.Int32	n	number of columns of dense matrices B and C.
System.Int32	k	number of columns of sparse matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	B	array of dimensions (ldb, n).
System.Int32	ldb	leading dimension of B. It must be at least max (1, k) if op(A) = A, and at least max (1, m) otherwise.
System.Single	beta	scalar used for multiplication. If beta is zero, C does not have to be a valid input.
CudaDeviceVariable<System.Single>	C	array of dimensions (ldc, n).
System.Int32	ldc	leading dimension of C. It must be at least max (1, m) if op(A) = A and at least max (1, k) otherwise.

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, cusparseMatDescr, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, CudaDeviceVariable<cuDoubleComplex> alpha, cusparseMatDescr descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<cuDoubleComplex>	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<cuDoubleComplex>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<cuDoubleComplex>	B
System.Int32	ldb
CudaDeviceVariable<cuDoubleComplex>	beta
CudaDeviceVariable<cuDoubleComplex>	C
System.Int32	ldc

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, cusparseMatDescr, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, CudaDeviceVariable<cuFloatComplex> alpha, cusparseMatDescr descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<cuFloatComplex>	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<cuFloatComplex>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<cuFloatComplex>	B
System.Int32	ldb
CudaDeviceVariable<cuFloatComplex>	beta
CudaDeviceVariable<cuFloatComplex>	C
System.Int32	ldc

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<Double>, cusparseMatDescr, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, CudaDeviceVariable<double> alpha, cusparseMatDescr descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<System.Double>	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Double>	B
System.Int32	ldb
CudaDeviceVariable<System.Double>	beta
CudaDeviceVariable<System.Double>	C
System.Int32	ldc

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, CudaDeviceVariable<Single>, cusparseMatDescr, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, CudaDeviceVariable<float> alpha, cusparseMatDescr descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<System.Single>	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Single>	B
System.Int32	ldb
CudaDeviceVariable<System.Single>	beta
CudaDeviceVariable<System.Single>	C
System.Int32	ldc

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, ref cuDoubleComplex, cusparseMatDescr, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, Int32, ref cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, ref cuDoubleComplex alpha, cusparseMatDescr descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> B, int ldb, ref cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
cuDoubleComplex	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<cuDoubleComplex>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<cuDoubleComplex>	B
System.Int32	ldb
cuDoubleComplex	beta
CudaDeviceVariable<cuDoubleComplex>	C
System.Int32	ldc

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, ref cuFloatComplex, cusparseMatDescr, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, Int32, ref cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, ref cuFloatComplex alpha, cusparseMatDescr descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> B, int ldb, ref cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
cuFloatComplex	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<cuFloatComplex>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<cuFloatComplex>	B
System.Int32	ldb
cuFloatComplex	beta
CudaDeviceVariable<cuFloatComplex>	C
System.Int32	ldc

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, ref Double, cusparseMatDescr, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Int32, ref Double, CudaDeviceVariable<Double>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, ref double alpha, cusparseMatDescr descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> B, int ldb, ref double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
System.Double	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Double>	B
System.Int32	ldb
System.Double	beta
CudaDeviceVariable<System.Double>	C
System.Int32	ldc

Csrmm2(cusparseOperation, cusparseOperation, Int32, Int32, Int32, Int32, ref Single, cusparseMatDescr, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Int32, ref Single, CudaDeviceVariable<Single>, Int32)

Declaration

public void Csrmm2(cusparseOperation transa, cusparseOperation transb, int m, int n, int k, int nnz, ref float alpha, cusparseMatDescr descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> B, int ldb, ref float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
cusparseOperation	transa
cusparseOperation	transb
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
System.Single	alpha
cusparseMatDescr	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Single>	B
System.Int32	ldb
System.Single	beta
CudaDeviceVariable<System.Single>	C
System.Int32	ldc

Csrmv(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

Csrmv(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

Csrmv(cusparseOperation, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> x, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

Csrmv(cusparseOperation, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> x, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

Csrmv(cusparseOperation, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> x, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
cuDoubleComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

Csrmv(cusparseOperation, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> x, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
cuFloatComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

Csrmv(cusparseOperation, Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Double, CudaDeviceVariable<Double>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> x, double beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
System.Double	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

Csrmv(cusparseOperation, Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Single, CudaDeviceVariable<Single>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void Csrmv(cusparseOperation transA, int m, int n, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> x, float beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
System.Single	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> x, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> x, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuDoubleComplex> x, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
cuDoubleComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<cuFloatComplex> x, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
cuFloatComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Double, CudaDeviceVariable<Double>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> x, double beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
System.Double	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrmvMP(cusparseOperation, Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Single, CudaDeviceVariable<Single>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in CSR storage format, x and y are dense vectors. Using a Merge Path load-balancing implementation.

Declaration

public void CsrmvMP(cusparseOperation transA, int m, int n, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> x, float beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix types are CUSPARSE_MATRIX_TYPE_GENERAL, CUSPARSE_MATRIX_TYPE_SYMMETRIC, and CUSPARSE_MATRIX_TYPE_HERMITIAN. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	x	vector of n elements if op(A) = A, and m elements if op(A) = AT or op(A) = AH.
System.Single	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of m elements if op(A) = A and n elements if op(A) = AT or op(A) = AH.

CsrsmAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsmAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsmAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsmAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsmSolve(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuDoubleComplex> x, int ldx, CudaDeviceVariable<cuDoubleComplex> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<cuDoubleComplex>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

CsrsmSolve(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuFloatComplex> x, int ldx, CudaDeviceVariable<cuFloatComplex> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<cuFloatComplex>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

CsrsmSolve(cusparseOperation, Int32, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<double> x, int ldx, CudaDeviceVariable<double> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<System.Double>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<System.Double>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

CsrsmSolve(cusparseOperation, Int32, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<float> x, int ldx, CudaDeviceVariable<float> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<System.Single>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<System.Single>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

CsrsmSolve(cusparseOperation, Int32, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuDoubleComplex> x, int ldx, CudaDeviceVariable<cuDoubleComplex> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<cuDoubleComplex>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

CsrsmSolve(cusparseOperation, Int32, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuFloatComplex> x, int ldx, CudaDeviceVariable<cuFloatComplex> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<cuFloatComplex>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

CsrsmSolve(cusparseOperation, Int32, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<double> x, int ldx, CudaDeviceVariable<double> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<System.Double>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<System.Double>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

CsrsmSolve(cusparseOperation, Int32, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

Solution of triangular linear system op(A) * Y = alpha * X, with multiple right-hand-sides, where A is a sparse matrix in CSR storage format, X and Y are dense and usually tall matrices.

Declaration

public void CsrsmSolve(cusparseOperation transA, int m, int n, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<float> x, int ldx, CudaDeviceVariable<float> y, int ldy)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows and columns of matrix A.
System.Int32	n	number of columns of matrix X and Y .
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure with information collected during the analysis phase (that should have been passed to the solve phase unchanged).
CudaDeviceVariable<System.Single>	x	right-hand-side array of dimensions (ldx, n).
System.Int32	ldx	leading dimension of X (that is >= max(1;m)).
CudaDeviceVariable<System.Single>	y	solution array of dimensions (ldy, n).
System.Int32	ldy	leading dimension of Y (that is >= max(1;m)).

Csrsort(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

This function sorts CSR format. The stable sorting is in-place.

The matrix type is regarded as CUSPARSE_MATRIX_TYPE_GENERAL implicitly. In other words, any symmetric property is ignored.

This function csrsort() requires buffer size returned by csrsort_bufferSizeExt().

The address of pBuffer must be multiple of 128 bytes. If not, CUSPARSE_STATUS_INVALID_VALUE is returned.

The parameter P is both input and output. If the user wants to compute sorted csrVal, P must be set as 0:1:(nnz-1) before csrsort(), and after csrsort(), new sorted value array satisfies csrVal_sorted = csrVal(P).

Declaration

public void Csrsort(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<int> P, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz unsorted column indices of A.
CudaDeviceVariable<System.Int32>	P	integer array of nnz sorted map indices.
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by CsrsortBufferSize().

CsrsortBufferSize(Int32, Int32, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function sorts CSR format. The stable sorting is in-place.

The matrix type is regarded as CUSPARSE_MATRIX_TYPE_GENERAL implicitly. In other words, any symmetric property is ignored.

This function csrsort() requires buffer size returned by csrsort_bufferSizeExt().

The address of pBuffer must be multiple of 128 bytes. If not, CUSPARSE_STATUS_INVALID_VALUE is returned.

The parameter P is both input and output. If the user wants to compute sorted csrVal, P must be set as 0:1:(nnz-1) before csrsort(), and after csrsort(), new sorted value array satisfies csrVal_sorted = csrVal(P).

Declaration

public SizeT CsrsortBufferSize(int m, int n, int nnz, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz unsorted column indices of A.

Returns

Type	Description
SizeT	number of bytes of the buffer.

Csrsv2Analysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Analysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrsv2Analysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Analysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrsv2Analysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Analysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrsv2Analysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, cusparseSolvePolicy, CudaDeviceVariable<Byte>)

This function performs the analysis phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Analysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().

Csrsv2BufferSize(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info)

This function returns the size of the buffer used in csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Csrsv2BufferSize(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrsv2BufferSize(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info)

This function returns the size of the buffer used in csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Csrsv2BufferSize(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrsv2BufferSize(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info)

This function returns the size of the buffer used in csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Csrsv2BufferSize(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrsv2BufferSize(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info)

This function returns the size of the buffer used in csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public SizeT Csrsv2BufferSize(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.

Returns

Type	Description
SizeT

Csrsv2Solve(cusparseOperation, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<cuDoubleComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuDoubleComplex>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<cuDoubleComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

Csrsv2Solve(cusparseOperation, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<cuFloatComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuFloatComplex>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<cuFloatComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

Csrsv2Solve(cusparseOperation, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<Double>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Double>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<double> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<System.Double>	y	solution vector of size m.

Csrsv2Solve(cusparseOperation, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<Single>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Single>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<float> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<System.Single>	y	solution vector of size m.

Csrsv2Solve(cusparseOperation, Int32, ref cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<cuDoubleComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuDoubleComplex>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, ref cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<cuDoubleComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

Csrsv2Solve(cusparseOperation, Int32, ref cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<cuFloatComplex>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<cuFloatComplex>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, ref cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<cuFloatComplex> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

Csrsv2Solve(cusparseOperation, Int32, ref Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<Double>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Double>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, ref double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<double> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<System.Double>	y	solution vector of size m.

Csrsv2Solve(cusparseOperation, Int32, ref Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsrsv2Info, CudaDeviceVariable<Single>, cusparseSolvePolicy, CudaDeviceVariable<Byte>, CudaDeviceVariable<Single>)

This function performs the solve phase of csrsv2, a new sparse triangular linear system op(A)*y = x.

Declaration

public void Csrsv2Solve(cusparseOperation transA, int m, ref float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseCsrsv2Info info, CudaDeviceVariable<float> x, cusparseSolvePolicy policy, CudaDeviceVariable<byte> buffer, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseCsrsv2Info	info	record of internal states based on different algorithms.
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
cusparseSolvePolicy	policy	The supported policies are CUSPARSE_SOLVE_POLICY_NO_LEVEL and CUSPARSE_SOLVE_POLICY_USE_LEVEL.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is returned by csrsv2_bufferSizeExt().
CudaDeviceVariable<System.Single>	y	solution vector of size m.

Csrsv2ZeroPivot(CudaSparseCsrsv2Info, CudaDeviceVariable<Int32>)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXcsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Csrsv2ZeroPivot(CudaSparseCsrsv2Info info, CudaDeviceVariable<int> position)

Parameters

Type	Name	Description
CudaSparseCsrsv2Info	info	info contains structural zero or numerical zero if the user already called csrsv2_analysis() or csrsv2_solve().
CudaDeviceVariable<System.Int32>	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Csrsv2ZeroPivot(CudaSparseCsrsv2Info, ref Int32)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) has either a structural zero or a numerical zero. Otherwise position=-1.

The position can be 0-based or 1-based, the same as the matrix.

Function cusparseXcsrsv2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Csrsv2ZeroPivot(CudaSparseCsrsv2Info info, ref int position)

Parameters

Type	Name	Description
CudaSparseCsrsv2Info	info	info contains structural zero or numerical zero if the user already called csrsv2_analysis() or csrsv2_solve().
System.Int32	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

CsrsvAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsvAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsvAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsvAnalysis(cusparseOperation, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvAnalysis(cusparseOperation transA, int m, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

CsrsvSolve(cusparseOperation, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

CsrsvSolve(cusparseOperation, Int32, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

CsrsvSolve(cusparseOperation, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<double> x, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Double>	y	solution vector of size m.

CsrsvSolve(cusparseOperation, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<float> x, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Single>	y	solution vector of size m.

CsrsvSolve(cusparseOperation, Int32, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

CsrsvSolve(cusparseOperation, Int32, cuFloatComplex, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

CsrsvSolve(cusparseOperation, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, double alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<double> x, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Double>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Double>	y	solution vector of size m.

CsrsvSolve(cusparseOperation, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in CSR storage format, x and y are dense vectors.

Declaration

public void CsrsvSolve(cusparseOperation transA, int m, float alpha, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<float> x, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal types CUSPARSE_DIAG_TYPE_UNIT and CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaDeviceVariable<System.Single>	csrValA	array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A. Length of csrColIndA gives the number nzz passed to CUSPARSE.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Single>	y	solution vector of size m.

Csru2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csru2csr(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csru2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csru2csr(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csru2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csru2csr(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csru2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo, CudaDeviceVariable<Byte>)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public void Csru2csr(int m, int n, int nnz, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL, Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().
CudaDeviceVariable<System.Byte>	pBuffer	buffer allocated by the user; the size is returned by Csru2csrBufferSize().

Csru2csrBufferSize(Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public SizeT Csru2csrBufferSize(int m, int n, int nnz, CudaDeviceVariable<cuDoubleComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<cuDoubleComplex>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().

Returns

Type	Description
SizeT	number of bytes of the buffer.

Csru2csrBufferSize(Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public SizeT Csru2csrBufferSize(int m, int n, int nnz, CudaDeviceVariable<cuFloatComplex> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<cuFloatComplex>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().

Returns

Type	Description
SizeT	number of bytes of the buffer.

Csru2csrBufferSize(Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public SizeT Csru2csrBufferSize(int m, int n, int nnz, CudaDeviceVariable<double> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<System.Double>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().

Returns

Type	Description
SizeT	number of bytes of the buffer.

Csru2csrBufferSize(Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparseCsru2csrInfo)

This function transfers unsorted CSR format to CSR format, and vice versa. The operation is in-place.

Declaration

public SizeT Csru2csrBufferSize(int m, int n, int nnz, CudaDeviceVariable<float> csrVal, CudaDeviceVariable<int> csrRowPtr, CudaDeviceVariable<int> csrColInd, CudaSparseCsru2csrInfo info)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	nnz	number of nonzero elements of matrix A.
CudaDeviceVariable<System.Single>	csrVal	array of nnz unsorted nonzero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	integer array of m+1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColInd	integer array of nnz unsorted column indices of A.
CudaSparseCsru2csrInfo	info	opaque structure initialized using cusparseCreateCsru2csrInfo().

Returns

Type	Description
SizeT	number of bytes of the buffer.

Dense2csc(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSC storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csc(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<int> nnzPerCol, CudaDeviceVariable<cuDoubleComplex> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerCol	array of size n containing the number of non-zero elements per column.
CudaDeviceVariable<cuDoubleComplex>	cscValA	Output: array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.

Dense2csc(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSC storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csc(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<int> nnzPerCol, CudaDeviceVariable<cuFloatComplex> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerCol	array of size n containing the number of non-zero elements per column.
CudaDeviceVariable<cuFloatComplex>	cscValA	Output: array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.

Dense2csc(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSC storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csc(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<int> nnzPerCol, CudaDeviceVariable<double> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerCol	array of size n containing the number of non-zero elements per column.
CudaDeviceVariable<System.Double>	cscValA	Output: array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.

Dense2csc(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSC storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csc(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<int> nnzPerCol, CudaDeviceVariable<float> cscValA, CudaDeviceVariable<int> cscRowIndA, CudaDeviceVariable<int> cscColPtrA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerCol	array of size n containing the number of non-zero elements per column.
CudaDeviceVariable<System.Single>	cscValA	Output: array of nnz (= cscRowPtrA(m)-cscRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscRowIndA	Output: integer array of nnz (= cscRowPtrA(m) - cscRowPtrA(0)) column indices of the non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	cscColPtrA	Output: integer array of n+1 elements that contains the start of every column and the end of the last column plus one.

Dense2csr(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSR storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csr(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaDeviceVariable<cuDoubleComplex>	csrValA	Output: array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	Output: integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	Output: integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.

Dense2csr(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSR storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csr(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaDeviceVariable<cuFloatComplex>	csrValA	Output: array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	Output: integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	Output: integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.

Dense2csr(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSR storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csr(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaDeviceVariable<System.Double>	csrValA	Output: array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	Output: integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	Output: integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.

Dense2csr(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine converts a dense matrix to a sparse matrix in the CSR storage format, using the information computed by the nnz routine.

Declaration

public void Dense2csr(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaDeviceVariable<System.Single>	csrValA	Output: array of nnz (= csrRowPtrA(m)-csrRowPtrA(0)) non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtrA	Output: integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	Output: integer array of nnz (= csrRowPtrA(m) - csrRowPtrA(0)) column indices of the non-zero elements of matrix A.

Dense2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a dense matrix to a sparse matrix in HYB storage format.

Declaration

public void Dense2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of the dense matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaSparseHybMat	hybA	Output: the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Dense2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a dense matrix to a sparse matrix in HYB storage format.

Declaration

public void Dense2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of the dense matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaSparseHybMat	hybA	Output: the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Dense2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a dense matrix to a sparse matrix in HYB storage format.

Declaration

public void Dense2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of the dense matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaDeviceVariable<System.Double>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaSparseHybMat	hybA	Output: the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Dense2hyb(Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Int32>, CudaSparseHybMat, Int32, cusparseHybPartition)

This routine converts a dense matrix to a sparse matrix in HYB storage format.

Declaration

public void Dense2hyb(int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<int> nnzPerRow, CudaSparseHybMat hybA, int userEllWidth, cusparseHybPartition partitionType)

Parameters

Type	Name	Description
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of the dense matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaDeviceVariable<System.Single>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRow	array of size m containing the number of non-zero elements per row.
CudaSparseHybMat	hybA	Output: the matrix A in HYB storage format.
System.Int32	userEllWidth	width of the regular (ELL) part of the matrix in HYB format, which should be less than maximum number of non-zeros per row and is only required if partitionType == CUSPARSE_HYB_PARTITION_USER.
cusparseHybPartition	partitionType	partitioning method to be used in the conversion (please refer to cusparseHybPartition_t on page 15 for details).

Dispose()

Dispose

Declaration

public void Dispose()

Dispose(Boolean)

For IDisposable

Declaration

protected virtual void Dispose(bool fDisposing)

Parameters

Type	Name	Description
System.Boolean	fDisposing

Dotci(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, cusparseIndexBase)

dot product of complex conjugate of a sparse vector x and a dense vector y.

Declaration

public void Dotci(CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> y, CudaDeviceVariable<cuDoubleComplex> result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format.
CudaDeviceVariable<cuDoubleComplex>	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Dotci(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, ref cuDoubleComplex, cusparseIndexBase)

dot product of complex conjugate of a sparse vector x and a dense vector y.

Declaration

public void Dotci(CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> y, ref cuDoubleComplex result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format.
cuDoubleComplex	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Dotci(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, cusparseIndexBase)

dot product of complex conjugate of a sparse vector x and a dense vector y.

Declaration

public void Dotci(CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> y, CudaDeviceVariable<cuFloatComplex> result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format.
CudaDeviceVariable<cuFloatComplex>	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Dotci(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, ref cuFloatComplex, cusparseIndexBase)

dot product of complex conjugate of a sparse vector x and a dense vector y.

Declaration

public void Dotci(CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> y, ref cuFloatComplex result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format.
cuFloatComplex	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> y, CudaDeviceVariable<cuDoubleComplex> result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format.
CudaDeviceVariable<cuDoubleComplex>	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, ref cuDoubleComplex, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> y, ref cuDoubleComplex result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format.
cuDoubleComplex	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> y, CudaDeviceVariable<cuFloatComplex> result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format.
CudaDeviceVariable<cuFloatComplex>	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, ref cuFloatComplex, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> y, ref cuFloatComplex result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format.
cuFloatComplex	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> y, CudaDeviceVariable<double> result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	y	vector in dense format.
CudaDeviceVariable<System.Double>	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, ref Double, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> y, ref double result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	y	vector in dense format.
System.Double	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> y, CudaDeviceVariable<float> result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	y	vector in dense format.
CudaDeviceVariable<System.Single>	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Doti(CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, ref Single, cusparseIndexBase)

dot product of a sparse vector x and a dense vector y

Declaration

public void Doti(CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> y, ref float result, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	y	vector in dense format.
System.Single	result	pointer to the location of the result in the device or host memory.
cusparseIndexBase	idxBase	Index base.

Finalize()

For dispose

Declaration

protected void Finalize()

Gebsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Let m(=mbrowBlockDim) be number of rows of A and n(=nbcolBlockDim) be number of columns of A, then A and C are mn sparse matrices. General BSR format of A contains nnzb(=bsrRowPtrA[mb] - bsrRowPtrA[0]) non-zero blocks whereas sparse matrix A contains nnz(=nnzbrowBlockDim*colBockDim) elements. The user must allocate enough space for arrays csrRowPtrC, csrColIndC and csrValC. The requirements are

csrRowPtrC of m+1 elements,

csrValC of nnz elements, and

csrColIndC of nnz elements.

Declaration

public void Gebsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDim, int colBlockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	csrValC	array of nnz non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix C.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the nonzero elements of matrix C.

Gebsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Let m(=mbrowBlockDim) be number of rows of A and n(=nbcolBlockDim) be number of columns of A, then A and C are mn sparse matrices. General BSR format of A contains nnzb(=bsrRowPtrA[mb] - bsrRowPtrA[0]) non-zero blocks whereas sparse matrix A contains nnz(=nnzbrowBlockDim*colBockDim) elements. The user must allocate enough space for arrays csrRowPtrC, csrColIndC and csrValC. The requirements are

csrRowPtrC of m+1 elements,

csrValC of nnz elements, and

csrColIndC of nnz elements.

Declaration

public void Gebsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDim, int colBlockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	csrValC	array of nnz non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix C.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the nonzero elements of matrix C.

Gebsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Let m(=mbrowBlockDim) be number of rows of A and n(=nbcolBlockDim) be number of columns of A, then A and C are mn sparse matrices. General BSR format of A contains nnzb(=bsrRowPtrA[mb] - bsrRowPtrA[0]) non-zero blocks whereas sparse matrix A contains nnz(=nnzbrowBlockDim*colBockDim) elements. The user must allocate enough space for arrays csrRowPtrC, csrColIndC and csrValC. The requirements are

csrRowPtrC of m+1 elements,

csrValC of nnz elements, and

csrColIndC of nnz elements.

Declaration

public void Gebsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDim, int colBlockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	csrValC	array of nnz non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix C.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the nonzero elements of matrix C.

Gebsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Let m(=mbrowBlockDim) be number of rows of A and n(=nbcolBlockDim) be number of columns of A, then A and C are mn sparse matrices. General BSR format of A contains nnzb(=bsrRowPtrA[mb] - bsrRowPtrA[0]) non-zero blocks whereas sparse matrix A contains nnz(=nnzbrowBlockDim*colBockDim) elements. The user must allocate enough space for arrays csrRowPtrC, csrColIndC and csrValC. The requirements are

csrRowPtrC of m+1 elements,

csrValC of nnz elements, and

csrColIndC of nnz elements.

Declaration

public void Gebsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDim, int colBlockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix C.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the nonzero elements of matrix C.

Gebsr2csr(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in CSR format (that is defined by arrays csrValC, csrRowPtrC, and csrColIndC).

Let m(=mbrowBlockDim) be number of rows of A and n(=nbcolBlockDim) be number of columns of A, then A and C are mn sparse matrices. General BSR format of A contains nnzb(=bsrRowPtrA[mb] - bsrRowPtrA[0]) non-zero blocks whereas sparse matrix A contains nnz(=nnzbrowBlockDim*colBockDim) elements. The user must allocate enough space for arrays csrRowPtrC, csrColIndC and csrValC. The requirements are

csrRowPtrC of m+1 elements,

csrValC of nnz elements, and

csrColIndC of nnz elements.

Declaration

public void Gebsr2csr(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDim, int colBlockDim, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	csrValC	array of nnz non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	csrRowPtrC	integer array of m+1 elements that contains the start of every row and the end of the last row plus one of matrix C.
CudaDeviceVariable<System.Int32>	csrColIndC	integer array of nnz column indices of the nonzero elements of matrix C.

Gebsr2gebsc(Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function can be seen as the same as csr2csc when regarding each block of size rowBlockDim*colBlockDim as a scalar.

This sparsity pattern of result matrix can also be seen as the transpose of the original sparse matrix but memory layout of a block does not change.

The user must know the size of buffer required by gebsr2gebsc by calling gebsr2gebsc_bufferSizeExt, allocate the buffer and pass the buffer pointer to gebsr2gebsc.

Declaration

public void Gebsr2gebsc(int mb, int nb, int nnzb, CudaDeviceVariable<cuDoubleComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim, CudaDeviceVariable<cuDoubleComplex> bscVal, CudaDeviceVariable<int> bscRowInd, CudaDeviceVariable<int> bscColPtr, cusparseAction copyValues, cusparseIndexBase baseIdx, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaDeviceVariable<cuDoubleComplex>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaDeviceVariable<cuDoubleComplex>	bscVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	bscRowInd	integer array of nnzb row indices of the non-zero blocks of matrix A
CudaDeviceVariable<System.Int32>	bscColPtr	integer array of nb+1 elements that contains the start of every block column and the end of the last block column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	baseIdx	CUSPARSE_INDEX_BASE_ZERO or CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsc_bufferSizeExt.

Gebsr2gebsc(Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function can be seen as the same as csr2csc when regarding each block of size rowBlockDim*colBlockDim as a scalar.

This sparsity pattern of result matrix can also be seen as the transpose of the original sparse matrix but memory layout of a block does not change.

The user must know the size of buffer required by gebsr2gebsc by calling gebsr2gebsc_bufferSizeExt, allocate the buffer and pass the buffer pointer to gebsr2gebsc.

Declaration

public void Gebsr2gebsc(int mb, int nb, int nnzb, CudaDeviceVariable<cuFloatComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim, CudaDeviceVariable<cuFloatComplex> bscVal, CudaDeviceVariable<int> bscRowInd, CudaDeviceVariable<int> bscColPtr, cusparseAction copyValues, cusparseIndexBase baseIdx, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaDeviceVariable<cuFloatComplex>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaDeviceVariable<cuFloatComplex>	bscVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	bscRowInd	integer array of nnzb row indices of the non-zero blocks of matrix A
CudaDeviceVariable<System.Int32>	bscColPtr	integer array of nb+1 elements that contains the start of every block column and the end of the last block column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	baseIdx	CUSPARSE_INDEX_BASE_ZERO or CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsc_bufferSizeExt.

Gebsr2gebsc(Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function can be seen as the same as csr2csc when regarding each block of size rowBlockDim*colBlockDim as a scalar.

This sparsity pattern of result matrix can also be seen as the transpose of the original sparse matrix but memory layout of a block does not change.

The user must know the size of buffer required by gebsr2gebsc by calling gebsr2gebsc_bufferSizeExt, allocate the buffer and pass the buffer pointer to gebsr2gebsc.

Declaration

public void Gebsr2gebsc(int mb, int nb, int nnzb, CudaDeviceVariable<double> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim, CudaDeviceVariable<double> bscVal, CudaDeviceVariable<int> bscRowInd, CudaDeviceVariable<int> bscColPtr, cusparseAction copyValues, cusparseIndexBase baseIdx, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaDeviceVariable<System.Double>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaDeviceVariable<System.Double>	bscVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	bscRowInd	integer array of nnzb row indices of the non-zero blocks of matrix A
CudaDeviceVariable<System.Int32>	bscColPtr	integer array of nb+1 elements that contains the start of every block column and the end of the last block column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	baseIdx	CUSPARSE_INDEX_BASE_ZERO or CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsc_bufferSizeExt.

Gebsr2gebsc(Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cusparseAction, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function can be seen as the same as csr2csc when regarding each block of size rowBlockDim*colBlockDim as a scalar.

This sparsity pattern of result matrix can also be seen as the transpose of the original sparse matrix but memory layout of a block does not change.

The user must know the size of buffer required by gebsr2gebsc by calling gebsr2gebsc_bufferSizeExt, allocate the buffer and pass the buffer pointer to gebsr2gebsc.

Declaration

public void Gebsr2gebsc(int mb, int nb, int nnzb, CudaDeviceVariable<float> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim, CudaDeviceVariable<float> bscVal, CudaDeviceVariable<int> bscRowInd, CudaDeviceVariable<int> bscColPtr, cusparseAction copyValues, cusparseIndexBase baseIdx, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaDeviceVariable<System.Single>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.
CudaDeviceVariable<System.Single>	bscVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A. It is only filled-in if copyValues is set to CUSPARSE_ACTION_NUMERIC.
CudaDeviceVariable<System.Int32>	bscRowInd	integer array of nnzb row indices of the non-zero blocks of matrix A
CudaDeviceVariable<System.Int32>	bscColPtr	integer array of nb+1 elements that contains the start of every block column and the end of the last block column plus one.
cusparseAction	copyValues	CUSPARSE_ACTION_SYMBOLIC or CUSPARSE_ACTION_NUMERIC.
cusparseIndexBase	baseIdx	CUSPARSE_INDEX_BASE_ZERO or CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsc_bufferSizeExt.

Gebsr2gebscBufferSize(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsc.

Declaration

public SizeT Gebsr2gebscBufferSize(int mb, int nb, CudaDeviceVariable<cuDoubleComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaDeviceVariable<cuDoubleComplex>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.

Returns

Type	Description
SizeT	number of bytes of the buffer used in the gebsr2gebsc.

Gebsr2gebscBufferSize(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsc.

Declaration

public SizeT Gebsr2gebscBufferSize(int mb, int nb, CudaDeviceVariable<cuFloatComplex> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaDeviceVariable<cuFloatComplex>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.

Returns

Type	Description
SizeT	number of bytes of the buffer used in the gebsr2gebsc.

Gebsr2gebscBufferSize(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsc.

Declaration

public SizeT Gebsr2gebscBufferSize(int mb, int nb, CudaDeviceVariable<double> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaDeviceVariable<System.Double>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.

Returns

Type	Description
SizeT	number of bytes of the buffer used in the gebsr2gebsc.

Gebsr2gebscBufferSize(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsc.

Declaration

public SizeT Gebsr2gebscBufferSize(int mb, int nb, CudaDeviceVariable<float> bsrVal, CudaDeviceVariable<int> bsrRowPtr, CudaDeviceVariable<int> bsrColInd, int rowBlockDim, int colBlockDim)

Parameters

Type	Name	Description
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaDeviceVariable<System.Single>	bsrVal	array of nnzbrowBlockDimcolBlockDim non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtr	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one.
CudaDeviceVariable<System.Int32>	bsrColInd	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDim	number of rows within a block of A.
System.Int32	colBlockDim	number of columns within a block of A.

Returns

Type	Description
SizeT	number of bytes of the buffer used in the gebsr2gebsc.

Gebsr2gebsr(cusparseDirection, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in another general BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Gebsr2gebsr(cusparseDirection dirA, int mb, int nb, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuDoubleComplex> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDimC, int colBlockDimC, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValC	array of nnzcrowBlockDimCcolBlockDimC non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mc+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzc block column indices of the non-zero blocks of matrix C.
System.Int32	rowBlockDimC	number of rows within a block of C
System.Int32	colBlockDimC	number of columns within a block of C
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsr_bufferSizeExt.

Gebsr2gebsr(cusparseDirection, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in another general BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Gebsr2gebsr(cusparseDirection dirA, int mb, int nb, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<cuFloatComplex> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDimC, int colBlockDimC, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValC	array of nnzcrowBlockDimCcolBlockDimC non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mc+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzc block column indices of the non-zero blocks of matrix C.
System.Int32	rowBlockDimC	number of rows within a block of C
System.Int32	colBlockDimC	number of columns within a block of C
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsr_bufferSizeExt.

Gebsr2gebsr(cusparseDirection, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in another general BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Gebsr2gebsr(cusparseDirection dirA, int mb, int nb, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDimC, int colBlockDimC, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValC	array of nnzcrowBlockDimCcolBlockDimC non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mc+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzc block column indices of the non-zero blocks of matrix C.
System.Int32	rowBlockDimC	number of rows within a block of C
System.Int32	colBlockDimC	number of columns within a block of C
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsr_bufferSizeExt.

Gebsr2gebsr(cusparseDirection, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in another general BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Gebsr2gebsr(cusparseDirection dirA, int mb, int nb, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> bsrValC, CudaDeviceVariable<int> bsrRowPtrC, CudaDeviceVariable<int> bsrColIndC, int rowBlockDimC, int colBlockDimC, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValC	array of nnzcrowBlockDimCcolBlockDimC non-zero elements of matrix C.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mc+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
CudaDeviceVariable<System.Int32>	bsrColIndC	integer array of nnzc block column indices of the non-zero blocks of matrix C.
System.Int32	rowBlockDimC	number of rows within a block of C
System.Int32	colBlockDimC	number of columns within a block of C
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsr_bufferSizeExt.

Gebsr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsrNnz and gebsr2gebsr.

Declaration

public SizeT Gebsr2gebsrBufferSize(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, int rowBlockDimC, int colBlockDimC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
System.Int32	rowBlockDimC	number of rows within a block of C.
System.Int32	colBlockDimC	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Gebsr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsrNnz and gebsr2gebsr.

Declaration

public SizeT Gebsr2gebsrBufferSize(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, int rowBlockDimC, int colBlockDimC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
System.Int32	rowBlockDimC	number of rows within a block of C.
System.Int32	colBlockDimC	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Gebsr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsrNnz and gebsr2gebsr.

Declaration

public SizeT Gebsr2gebsrBufferSize(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, int rowBlockDimC, int colBlockDimC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
System.Int32	rowBlockDimC	number of rows within a block of C.
System.Int32	colBlockDimC	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Gebsr2gebsrBufferSize(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, Int32, Int32)

This function returns size of buffer used in computing gebsr2gebsrNnz and gebsr2gebsr.

Declaration

public SizeT Gebsr2gebsrBufferSize(cusparseDirection dirA, int mb, int nb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> bsrValA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, int rowBlockDimC, int colBlockDimC)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	bsrValA	array of nnzbrowBlockDimAcolBlockDimA non-zero elements of matrix A.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
System.Int32	rowBlockDimC	number of rows within a block of C.
System.Int32	colBlockDimC	number of columns within a block of C.

Returns

Type	Description
SizeT	number of bytes of the buffer used in csr2gebsrNnz() and csr2gebsr().

Gebsr2gebsrNnz(cusparseDirection, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, Int32, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in another general BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Gebsr2gebsrNnz(cusparseDirection dirA, int mb, int nb, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> bsrRowPtrC, int rowBlockDimC, int colBlockDimC, CudaDeviceVariable<int> nnzTotalDevHostPtr, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mc+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
System.Int32	rowBlockDimC	number of rows within a block of C
System.Int32	colBlockDimC	number of columns within a block of C
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr	total number of nonzero blocks of C. nnzTotalDevHostPtr is the same as bsrRowPtrC[mc]-bsrRowPtrC[0]
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsr_bufferSizeExt.

Gebsr2gebsrNnz(cusparseDirection, Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, Int32, Int32, ref Int32, CudaDeviceVariable<Byte>)

This function converts a sparse matrix in general BSR format (that is defined by the three arrays bsrValA, bsrRowPtrA, and bsrColIndA) into a sparse matrix in another general BSR format (that is defined by arrays bsrValC, bsrRowPtrC, and bsrColIndC).

Declaration

public void Gebsr2gebsrNnz(cusparseDirection dirA, int mb, int nb, int nnzb, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<int> bsrRowPtrA, CudaDeviceVariable<int> bsrColIndA, int rowBlockDimA, int colBlockDimA, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> bsrRowPtrC, int rowBlockDimC, int colBlockDimC, ref int nnzTotalDevHostPtr, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
cusparseDirection	dirA	storage format of blocks, either CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	mb	number of block rows of sparse matrix A.
System.Int32	nb	number of block columns of sparse matrix A.
System.Int32	nnzb	number of nonzero blocks of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	bsrRowPtrA	integer array of mb+1 elements that contains the start of every block row and the end of the last block row plus one of matrix A.
CudaDeviceVariable<System.Int32>	bsrColIndA	integer array of nnzb column indices of the nonzero blocks of matrix A.
System.Int32	rowBlockDimA	number of rows within a block of A.
System.Int32	colBlockDimA	number of columns within a block of A.
CudaSparseMatrixDescriptor	descrC	the descriptor of matrix C. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Int32>	bsrRowPtrC	integer array of mc+1 elements that contains the start of every block row and the end of the last block row plus one of matrix C.
System.Int32	rowBlockDimC	number of rows within a block of C
System.Int32	colBlockDimC	number of columns within a block of C
System.Int32	nnzTotalDevHostPtr	total number of nonzero blocks of C. nnzTotalDevHostPtr is the same as bsrRowPtrC[mc]-bsrRowPtrC[0]
CudaDeviceVariable<System.Byte>	buffer	buffer allocated by the user, the size is return by gebsr2gebsr_bufferSizeExt.

Gemmi(Int32, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<cuDoubleComplex>	alpha
CudaDeviceVariable<cuDoubleComplex>	A
System.Int32	lda
CudaDeviceVariable<cuDoubleComplex>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
CudaDeviceVariable<cuDoubleComplex>	beta
CudaDeviceVariable<cuDoubleComplex>	C
System.Int32	ldc

Gemmi(Int32, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<cuFloatComplex>	alpha
CudaDeviceVariable<cuFloatComplex>	A
System.Int32	lda
CudaDeviceVariable<cuFloatComplex>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
CudaDeviceVariable<cuFloatComplex>	beta
CudaDeviceVariable<cuFloatComplex>	C
System.Int32	ldc

Gemmi(Int32, Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<System.Double>	alpha
CudaDeviceVariable<System.Double>	A
System.Int32	lda
CudaDeviceVariable<System.Double>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
CudaDeviceVariable<System.Double>	beta
CudaDeviceVariable<System.Double>	C
System.Int32	ldc

Gemmi(Int32, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
CudaDeviceVariable<System.Single>	alpha
CudaDeviceVariable<System.Single>	A
System.Int32	lda
CudaDeviceVariable<System.Single>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
CudaDeviceVariable<System.Single>	beta
CudaDeviceVariable<System.Single>	C
System.Int32	ldc

Gemmi(Int32, Int32, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
cuDoubleComplex	alpha
CudaDeviceVariable<cuDoubleComplex>	A
System.Int32	lda
CudaDeviceVariable<cuDoubleComplex>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
cuDoubleComplex	beta
CudaDeviceVariable<cuDoubleComplex>	C
System.Int32	ldc

Gemmi(Int32, Int32, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
cuFloatComplex	alpha
CudaDeviceVariable<cuFloatComplex>	A
System.Int32	lda
CudaDeviceVariable<cuFloatComplex>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
cuFloatComplex	beta
CudaDeviceVariable<cuFloatComplex>	C
System.Int32	ldc

Gemmi(Int32, Int32, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, CudaDeviceVariable<Double>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
System.Double	alpha
CudaDeviceVariable<System.Double>	A
System.Int32	lda
CudaDeviceVariable<System.Double>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
System.Double	beta
CudaDeviceVariable<System.Double>	C
System.Int32	ldc

Gemmi(Int32, Int32, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaDeviceVariable<Single>, Int32)

Description: dense - sparse matrix multiplication C = alpha * A * B + beta * C, where A is column-major dense matrix, B is a sparse matrix in CSC format, and C is column-major dense matrix.

Declaration

public void Gemmi(int m, int n, int k, int nnz, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> cscValB, CudaDeviceVariable<int> cscColPtrB, CudaDeviceVariable<int> cscRowIndB, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	k
System.Int32	nnz
System.Single	alpha
CudaDeviceVariable<System.Single>	A
System.Int32	lda
CudaDeviceVariable<System.Single>	cscValB
CudaDeviceVariable<System.Int32>	cscColPtrB
CudaDeviceVariable<System.Int32>	cscRowIndB
System.Single	beta
CudaDeviceVariable<System.Single>	C
System.Int32	ldc

Gemvi(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, int nnz, CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<cuDoubleComplex>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

Gemvi(cusparseOperation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, int nnz, CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<cuFloatComplex>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

Gemvi(cusparseOperation, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, int nnz, CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<System.Double>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

Gemvi(cusparseOperation, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, int nnz, CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<System.Single>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

Gemvi(cusparseOperation, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, int nnz, CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<cuDoubleComplex>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
cuDoubleComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

Gemvi(cusparseOperation, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, int nnz, CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<cuFloatComplex>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
cuFloatComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

Gemvi(cusparseOperation, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, Double, CudaDeviceVariable<Double>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, double alpha, CudaDeviceVariable<double> A, int lda, int nnz, CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, double beta, CudaDeviceVariable<double> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<System.Double>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
System.Double	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

Gemvi(cusparseOperation, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, Single, CudaDeviceVariable<Single>, cusparseIndexBase, CudaDeviceVariable<Byte>)

This function performs the matrix-vector operation

y = alpha * op(A) * x + B * y

A is an m x n dense matrix and a sparse vector x that is defined in a sparse storage format by the two arrays xVal, xInd of length nnz, and y is a dense vector; alpha and beta are scalars.

Declaration

public void Gemvi(cusparseOperation transA, int m, int n, float alpha, CudaDeviceVariable<float> A, int lda, int nnz, CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, float beta, CudaDeviceVariable<float> y, cusparseIndexBase idxBase, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	the pointer to dense matrix A.
System.Int32	lda	size of the leading dimension of A.
System.Int32	nnz	number of nonzero elements of vector x.
CudaDeviceVariable<System.Single>	xVal	sparse vector of nnz elements of size n if op(A) = A, and of size m if op(A) = A^T or op(A) = A^H
CudaDeviceVariable<System.Int32>	xInd	Indices of non-zero values in x
System.Single	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	dense vector of m elements if op(A) = A, and of n elements if op(A) = A^T or op(A) = A^H
cusparseIndexBase	idxBase	0 or 1, for 0 based or 1 based indexing, respectively
CudaDeviceVariable<System.Byte>	pBuffer	working space buffer, of size given by Xgemvi_getBufferSize()

GemviCBufferSize(cusparseOperation, Int32, Int32, Int32)

This function returns size of buffer used in gemvi(). A is an (m)x(n) dense matrix.

Declaration

public int GemviCBufferSize(cusparseOperation transA, int m, int n, int nnz)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix Y.
System.Int32	nnz	number of nonzero entries of vector x multiplying A.

Returns

Type	Description
System.Int32	number of elements needed the buffer used in gemvi().

GemviDBufferSize(cusparseOperation, Int32, Int32, Int32)

This function returns size of buffer used in gemvi(). A is an (m)x(n) dense matrix.

Declaration

public int GemviDBufferSize(cusparseOperation transA, int m, int n, int nnz)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix Y.
System.Int32	nnz	number of nonzero entries of vector x multiplying A.

Returns

Type	Description
System.Int32	number of elements needed the buffer used in gemvi().

GemviSBufferSize(cusparseOperation, Int32, Int32, Int32)

This function returns size of buffer used in gemvi(). A is an (m)x(n) dense matrix.

Declaration

public int GemviSBufferSize(cusparseOperation transA, int m, int n, int nnz)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix Y.
System.Int32	nnz	number of nonzero entries of vector x multiplying A.

Returns

Type	Description
System.Int32	number of elements needed the buffer used in gemvi().

GemviZBufferSize(cusparseOperation, Int32, Int32, Int32)

This function returns size of buffer used in gemvi(). A is an (m)x(n) dense matrix.

Declaration

public int GemviZBufferSize(cusparseOperation transA, int m, int n, int nnz)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix Y.
System.Int32	nnz	number of nonzero entries of vector x multiplying A.

Returns

Type	Description
System.Int32	number of elements needed the buffer used in gemvi().

GetLevelInfo(CudaSparseSolveAnalysisInfo, out Int32, out CudaDeviceVariable<Int32>, out CudaDeviceVariable<Int32>)

This function returns the number of levels and the assignment of rows into the levels computed by either the csrsv_analysis, csrsm_analysis or hybsv_analysis routines.

Declaration

public void GetLevelInfo(CudaSparseSolveAnalysisInfo info, out int nLevels, out CudaDeviceVariable<int> levelPtr, out CudaDeviceVariable<int> levelIdx)

Parameters

Type	Name	Description
CudaSparseSolveAnalysisInfo	info	the pointer to the solve and analysis structure.
System.Int32	nLevels	number of levels.
CudaDeviceVariable<System.Int32>	levelPtr	integer array of nlevels+1 elements that contains the start of every level and the end of the last level plus one.
CudaDeviceVariable<System.Int32>	levelIdx	integer array of m (number of rows in the matrix) elements that contains the row indices belonging to every level.

GetPointerMode()

Returns the pointer mode for scalar values (host or device pointer)

Declaration

public cusparsePointerMode GetPointerMode()

Returns

Type	Description
cusparsePointerMode

GetStream()

Gets the cuda stream to use

Declaration

public CUstream GetStream()

Returns

Type	Description
CUstream

GetVersion()

Returns the version of the underlying CUSPARSE library

Declaration

public Version GetVersion()

Returns

Type	Description
System.Version

Gthr(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from dense vector y into sparse vector x.

Declaration

public void Gthr(CudaDeviceVariable<cuDoubleComplex> y, CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format (of size >= max(xInd)-idxBase+1).
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gthr(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from dense vector y into sparse vector x.

Declaration

public void Gthr(CudaDeviceVariable<cuFloatComplex> y, CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format (of size >= max(xInd)-idxBase+1).
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gthr(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from dense vector y into sparse vector x.

Declaration

public void Gthr(CudaDeviceVariable<double> y, CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	y	vector in dense format (of size >= max(xInd)-idxBase+1).
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gthr(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from dense vector y into sparse vector x.

Declaration

public void Gthr(CudaDeviceVariable<float> y, CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	y	vector in dense format (of size >= max(xInd)-idxBase+1).
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gthrz(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from desne vector y into sparse vector x (also replacing these elements in y by zeros).

Declaration

public void Gthrz(CudaDeviceVariable<cuDoubleComplex> y, CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format with elements indexed by xInd set to zero (it is unchanged if nnz == 0).
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gthrz(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from desne vector y into sparse vector x (also replacing these elements in y by zeros).

Declaration

public void Gthrz(CudaDeviceVariable<cuFloatComplex> y, CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format with elements indexed by xInd set to zero (it is unchanged if nnz == 0).
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gthrz(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from desne vector y into sparse vector x (also replacing these elements in y by zeros).

Declaration

public void Gthrz(CudaDeviceVariable<double> y, CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	y	vector in dense format with elements indexed by xInd set to zero (it is unchanged if nnz == 0).
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gthrz(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, cusparseIndexBase)

Gather of non-zero elements from desne vector y into sparse vector x (also replacing these elements in y by zeros).

Declaration

public void Gthrz(CudaDeviceVariable<float> y, CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	y	vector in dense format with elements indexed by xInd set to zero (it is unchanged if nnz == 0).
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values that were gathered from vector y (that is unchanged if nnz == 0).
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Gtsv(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv(int m, int n, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<cuDoubleComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<cuDoubleComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Gtsv(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv(int m, int n, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<cuFloatComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<cuFloatComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Gtsv(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv(int m, int n, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<System.Double>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Double>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<System.Double>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Double>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Gtsv(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv(int m, int n, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<System.Single>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Single>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<System.Single>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Single>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Gtsv_nopivot(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

Declaration

public void Gtsv_nopivot(int m, int n, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<cuDoubleComplex>	dl
CudaDeviceVariable<cuDoubleComplex>	d
CudaDeviceVariable<cuDoubleComplex>	du
CudaDeviceVariable<cuDoubleComplex>	B
System.Int32	ldb

Gtsv_nopivot(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

Declaration

public void Gtsv_nopivot(int m, int n, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<cuFloatComplex>	dl
CudaDeviceVariable<cuFloatComplex>	d
CudaDeviceVariable<cuFloatComplex>	du
CudaDeviceVariable<cuFloatComplex>	B
System.Int32	ldb

Gtsv_nopivot(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

Declaration

public void Gtsv_nopivot(int m, int n, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	dl
CudaDeviceVariable<System.Double>	d
CudaDeviceVariable<System.Double>	du
CudaDeviceVariable<System.Double>	B
System.Int32	ldb

Gtsv_nopivot(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

Declaration

public void Gtsv_nopivot(int m, int n, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	dl
CudaDeviceVariable<System.Single>	d
CudaDeviceVariable<System.Single>	du
CudaDeviceVariable<System.Single>	B
System.Int32	ldb

Gtsv2(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Byte>)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv2(int m, int n, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<cuDoubleComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<cuDoubleComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Byte>)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv2(int m, int n, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<cuFloatComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<cuFloatComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Byte>)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv2(int m, int n, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<System.Double>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Double>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<System.Double>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Double>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Byte>)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public void Gtsv2(int m, int n, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<System.Single>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Single>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<System.Single>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Single>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2_nopivot(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Byte>)

Declaration

public void Gtsv2_nopivot(int m, int n, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<cuDoubleComplex>	dl
CudaDeviceVariable<cuDoubleComplex>	d
CudaDeviceVariable<cuDoubleComplex>	du
CudaDeviceVariable<cuDoubleComplex>	B
System.Int32	ldb
CudaDeviceVariable<System.Byte>	buffer

Gtsv2_nopivot(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Byte>)

Declaration

public void Gtsv2_nopivot(int m, int n, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<cuFloatComplex>	dl
CudaDeviceVariable<cuFloatComplex>	d
CudaDeviceVariable<cuFloatComplex>	du
CudaDeviceVariable<cuFloatComplex>	B
System.Int32	ldb
CudaDeviceVariable<System.Byte>	buffer

Gtsv2_nopivot(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Byte>)

Declaration

public void Gtsv2_nopivot(int m, int n, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	dl
CudaDeviceVariable<System.Double>	d
CudaDeviceVariable<System.Double>	du
CudaDeviceVariable<System.Double>	B
System.Int32	ldb
CudaDeviceVariable<System.Byte>	buffer

Gtsv2_nopivot(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Byte>)

Declaration

public void Gtsv2_nopivot(int m, int n, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	dl
CudaDeviceVariable<System.Single>	d
CudaDeviceVariable<System.Single>	du
CudaDeviceVariable<System.Single>	B
System.Int32	ldb
CudaDeviceVariable<System.Byte>	buffer

Gtsv2_nopivotGetBufferSize(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

Declaration

public SizeT Gtsv2_nopivotGetBufferSize(int m, int n, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<cuDoubleComplex>	dl
CudaDeviceVariable<cuDoubleComplex>	d
CudaDeviceVariable<cuDoubleComplex>	du
CudaDeviceVariable<cuDoubleComplex>	B
System.Int32	ldb

Returns

Type	Description
SizeT

Gtsv2_nopivotGetBufferSize(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

Declaration

public SizeT Gtsv2_nopivotGetBufferSize(int m, int n, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<cuFloatComplex>	dl
CudaDeviceVariable<cuFloatComplex>	d
CudaDeviceVariable<cuFloatComplex>	du
CudaDeviceVariable<cuFloatComplex>	B
System.Int32	ldb

Returns

Type	Description
SizeT

Gtsv2_nopivotGetBufferSize(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

Declaration

public SizeT Gtsv2_nopivotGetBufferSize(int m, int n, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	dl
CudaDeviceVariable<System.Double>	d
CudaDeviceVariable<System.Double>	du
CudaDeviceVariable<System.Double>	B
System.Int32	ldb

Returns

Type	Description
SizeT

Gtsv2_nopivotGetBufferSize(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

Declaration

public SizeT Gtsv2_nopivotGetBufferSize(int m, int n, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	dl
CudaDeviceVariable<System.Single>	d
CudaDeviceVariable<System.Single>	du
CudaDeviceVariable<System.Single>	B
System.Int32	ldb

Returns

Type	Description
SizeT

Gtsv2GetBufferSize(Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public SizeT Gtsv2GetBufferSize(int m, int n, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<cuDoubleComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<cuDoubleComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Returns

Type	Description
SizeT

Gtsv2GetBufferSize(Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public SizeT Gtsv2GetBufferSize(int m, int n, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<cuFloatComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<cuFloatComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Returns

Type	Description
SizeT

Gtsv2GetBufferSize(Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public SizeT Gtsv2GetBufferSize(int m, int n, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<System.Double>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Double>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<System.Double>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Double>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Returns

Type	Description
SizeT

Gtsv2GetBufferSize(Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

Solution of tridiagonal linear system A * B = B, with multiple right-hand-sides. The coefficient matrix A is composed of lower (dl), main (d) and upper (du) diagonals, and the right-hand-sides B are overwritten with the solution.

Declaration

public SizeT Gtsv2GetBufferSize(int m, int n, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> B, int ldb)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
System.Int32	n	number of right-hand-sides, columns of matrix B.
CudaDeviceVariable<System.Single>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Single>	d	dense array containing the main diagonal of the tri-diagonal linear system.
CudaDeviceVariable<System.Single>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Single>	B	dense right-hand-side array of dimensions (ldb, m).
System.Int32	ldb	leading dimension of B (that is >= max(1;m)).

Returns

Type	Description
SizeT

Gtsv2StridedBatch(Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, Int32, CudaDeviceVariable<Byte>)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void Gtsv2StridedBatch(int m, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> x, int batchCount, int batchStride, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<cuDoubleComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<cuDoubleComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2StridedBatch(Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, Int32, CudaDeviceVariable<Byte>)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void Gtsv2StridedBatch(int m, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> x, int batchCount, int batchStride, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<cuFloatComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<cuFloatComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2StridedBatch(Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, Int32, CudaDeviceVariable<Byte>)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void Gtsv2StridedBatch(int m, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> x, int batchCount, int batchStride, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<System.Double>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Double>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<System.Double>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Double>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2StridedBatch(Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, Int32, CudaDeviceVariable<Byte>)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void Gtsv2StridedBatch(int m, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> x, int batchCount, int batchStride, CudaDeviceVariable<byte> buffer)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<System.Single>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Single>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<System.Single>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Single>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).
CudaDeviceVariable<System.Byte>	buffer	Buffer

Gtsv2StridedBatchGetBufferSize(Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public SizeT Gtsv2StridedBatchGetBufferSize(int m, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<cuDoubleComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<cuDoubleComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

Returns

Type	Description
SizeT

Gtsv2StridedBatchGetBufferSize(Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public SizeT Gtsv2StridedBatchGetBufferSize(int m, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<cuFloatComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<cuFloatComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

Returns

Type	Description
SizeT

Gtsv2StridedBatchGetBufferSize(Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public SizeT Gtsv2StridedBatchGetBufferSize(int m, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<System.Double>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Double>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<System.Double>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Double>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

Returns

Type	Description
SizeT

Gtsv2StridedBatchGetBufferSize(Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public SizeT Gtsv2StridedBatchGetBufferSize(int m, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<System.Single>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Single>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<System.Single>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Single>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

Returns

Type	Description
SizeT

GtsvStridedBatch(Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void GtsvStridedBatch(int m, CudaDeviceVariable<cuDoubleComplex> dl, CudaDeviceVariable<cuDoubleComplex> d, CudaDeviceVariable<cuDoubleComplex> du, CudaDeviceVariable<cuDoubleComplex> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<cuDoubleComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<cuDoubleComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<cuDoubleComplex>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

GtsvStridedBatch(Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void GtsvStridedBatch(int m, CudaDeviceVariable<cuFloatComplex> dl, CudaDeviceVariable<cuFloatComplex> d, CudaDeviceVariable<cuFloatComplex> du, CudaDeviceVariable<cuFloatComplex> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<cuFloatComplex>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<cuFloatComplex>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<cuFloatComplex>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

GtsvStridedBatch(Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void GtsvStridedBatch(int m, CudaDeviceVariable<double> dl, CudaDeviceVariable<double> d, CudaDeviceVariable<double> du, CudaDeviceVariable<double> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<System.Double>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Double>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<System.Double>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Double>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

GtsvStridedBatch(Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, Int32)

Solution of a set of tridiagonal linear systems A * x = x, each with a single right-hand-side. The coefficient matrices A are composed of lower (dl), main (d) and upper (du) diagonals and stored separated by a batchStride, while the right-hand-sides x are also separated by a batchStride.

Declaration

public void GtsvStridedBatch(int m, CudaDeviceVariable<float> dl, CudaDeviceVariable<float> d, CudaDeviceVariable<float> du, CudaDeviceVariable<float> x, int batchCount, int batchStride)

Parameters

Type	Name	Description
System.Int32	m	the size of the linear system (must be >= 3).
CudaDeviceVariable<System.Single>	dl	dense array containing the lower diagonal of the tri-diagonal linear system. The lower diagonal dl(i) that corresponds to the ith linear system starts at location dl + batchStride * i in memory. Also, the first element of each lower diagonal must be zero.
CudaDeviceVariable<System.Single>	d	dense array containing the main diagonal of the tri-diagonal linear system. The main diagonal d(i) that corresponds to the ith linear system starts at location d + batchStride * i in memory.
CudaDeviceVariable<System.Single>	du	dense array containing the upper diagonal of the tri-diagonal linear system. The upper diagonal du(i) that corresponds to the ith linear system starts at location du + batchStride * i in memory. Also, the last element of each upper diagonal must be zero.
CudaDeviceVariable<System.Single>	x	dense array that contains the right-hand-side of the tridiagonal linear system. The right-hand-side x(i) that corresponds to the ith linear system starts at location x + batchStride * i in memory.
System.Int32	batchCount	Number of systems to solve.
System.Int32	batchStride	stride (number of elements) that separates the vectors of every system (must be at least m).

Hyb2csc(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Declaration

public void Hyb2csc(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuDoubleComplex> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA
CudaSparseHybMat	hybA
CudaDeviceVariable<cuDoubleComplex>	cscVal
CudaDeviceVariable<System.Int32>	cscRowInd
CudaDeviceVariable<System.Int32>	cscColPtr

Hyb2csc(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Declaration

public void Hyb2csc(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuFloatComplex> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA
CudaSparseHybMat	hybA
CudaDeviceVariable<cuFloatComplex>	cscVal
CudaDeviceVariable<System.Int32>	cscRowInd
CudaDeviceVariable<System.Int32>	cscColPtr

Hyb2csc(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Declaration

public void Hyb2csc(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<double> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA
CudaSparseHybMat	hybA
CudaDeviceVariable<System.Double>	cscVal
CudaDeviceVariable<System.Int32>	cscRowInd
CudaDeviceVariable<System.Int32>	cscColPtr

Hyb2csc(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Declaration

public void Hyb2csc(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<float> cscVal, CudaDeviceVariable<int> cscRowInd, CudaDeviceVariable<int> cscColPtr)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA
CudaSparseHybMat	hybA
CudaDeviceVariable<System.Single>	cscVal
CudaDeviceVariable<System.Int32>	cscRowInd
CudaDeviceVariable<System.Int32>	cscColPtr

Hyb2csr(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in HYB format into a sparse matrix in CSR format.

This function requires some amount of temporary storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Hyb2csr(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuDoubleComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format
CudaDeviceVariable<cuDoubleComplex>	csrValA	array of nnz csrRowPtrA(m) csrRowPtrA(0) non-zero elements of matrix A
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every column and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix .

Hyb2csr(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in HYB format into a sparse matrix in CSR format.

This function requires some amount of temporary storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Hyb2csr(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuFloatComplex> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format
CudaDeviceVariable<cuFloatComplex>	csrValA	array of nnz csrRowPtrA(m) csrRowPtrA(0) non-zero elements of matrix A
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every column and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix .

Hyb2csr(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in HYB format into a sparse matrix in CSR format.

This function requires some amount of temporary storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Hyb2csr(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format
CudaDeviceVariable<System.Double>	csrValA	array of nnz csrRowPtrA(m) csrRowPtrA(0) non-zero elements of matrix A
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every column and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix .

Hyb2csr(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This function converts a sparse matrix in HYB format into a sparse matrix in CSR format.

This function requires some amount of temporary storage. It is executed asynchronously with respect to the host and it may return control to the application on the host before the result is ready.

Declaration

public void Hyb2csr(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format
CudaDeviceVariable<System.Single>	csrValA	array of nnz csrRowPtrA(m) csrRowPtrA(0) non-zero elements of matrix A
CudaDeviceVariable<System.Int32>	csrRowPtrA	integer array of m+1 elements that contains the start of every column and the end of the last row plus one.
CudaDeviceVariable<System.Int32>	csrColIndA	integer array of nnz csrRowPtrA(m) csrRowPtrA(0) column indices of the nonzero elements of matrix .

Hyb2dense(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuDoubleComplex>, Int32)

This routine converts a sparse matrix in HYB storage format to a dense matrix.

Declaration

public void Hyb2dense(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of the matrix A in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	the matrix A in HYB storage format.

Hyb2dense(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuFloatComplex>, Int32)

This routine converts a sparse matrix in HYB storage format to a dense matrix.

Declaration

public void Hyb2dense(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of the matrix A in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	the matrix A in HYB storage format.

Hyb2dense(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Double>, Int32)

This routine converts a sparse matrix in HYB storage format to a dense matrix.

Declaration

public void Hyb2dense(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of the matrix A in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<System.Double>	A	array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	the matrix A in HYB storage format.

Hyb2dense(CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Single>, Int32)

This routine converts a sparse matrix in HYB storage format to a dense matrix.

Declaration

public void Hyb2dense(CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
CudaSparseMatrixDescriptor	descrA	the descriptor of the matrix A in Hyb format. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<System.Single>	A	array of dimensions (lda, n) that is filled in with the values of the sparse matrix.
System.Int32	lda	the matrix A in HYB storage format.

Hybmv(cusparseOperation, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<cuDoubleComplex>	x	vector of n elements.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of m elements.

Hybmv(cusparseOperation, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<cuFloatComplex>	x	vector of n elements.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of m elements.

Hybmv(cusparseOperation, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<double> x, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<System.Double>	x	vector of n elements.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of m elements.

Hybmv(cusparseOperation, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<float> x, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<System.Single>	x	vector of n elements.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of m elements.

Hybmv(cusparseOperation, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuDoubleComplex>, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuDoubleComplex> x, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<cuDoubleComplex>	x	vector of n elements.
cuDoubleComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector of m elements.

Hybmv(cusparseOperation, cuFloatComplex, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<cuFloatComplex>, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<cuFloatComplex> x, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<cuFloatComplex>	x	vector of n elements.
cuFloatComplex	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector of m elements.

Hybmv(cusparseOperation, Double, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Double>, Double, CudaDeviceVariable<Double>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, double alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<double> x, double beta, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<System.Double>	x	vector of n elements.
System.Double	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector of m elements.

Hybmv(cusparseOperation, Single, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaDeviceVariable<Single>, Single, CudaDeviceVariable<Single>)

Matrix-vector multiplication y = alpha * op(A) * x + beta * y, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void Hybmv(cusparseOperation transA, float alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaDeviceVariable<float> x, float beta, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A).
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaDeviceVariable<System.Single>	x	vector of n elements.
System.Single	beta	scalar used for multiplication. If beta is zero, y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector of m elements.

HybsvAnalysis<T>(cusparseOperation, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvAnalysis<T>(cusparseOperation transA, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info)
    where T : struct

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).

Type Parameters

Name	Description
T	data type: float, double, cuFloatComplex or cuDoubleComplex

HybsvSolve(cusparseOperation, CudaDeviceVariable<cuDoubleComplex>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, CudaDeviceVariable<cuDoubleComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

HybsvSolve(cusparseOperation, CudaDeviceVariable<cuFloatComplex>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, CudaDeviceVariable<cuFloatComplex> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

HybsvSolve(cusparseOperation, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, CudaDeviceVariable<double> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<double> x, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Double>	y	solution vector of size m.

HybsvSolve(cusparseOperation, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, CudaDeviceVariable<float> alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<float> x, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Single>	y	solution vector of size m.

HybsvSolve(cusparseOperation, cuDoubleComplex, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, cuDoubleComplex alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuDoubleComplex> x, CudaDeviceVariable<cuDoubleComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuDoubleComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuDoubleComplex>	y	solution vector of size m.

HybsvSolve(cusparseOperation, cuFloatComplex, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, cuFloatComplex alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<cuFloatComplex> x, CudaDeviceVariable<cuFloatComplex> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
cuFloatComplex	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<cuFloatComplex>	x	right-hand-side vector of size m.
CudaDeviceVariable<cuFloatComplex>	y	solution vector of size m.

HybsvSolve(cusparseOperation, Double, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, double alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<double> x, CudaDeviceVariable<double> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
System.Double	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Double>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Double>	y	solution vector of size m.

HybsvSolve(cusparseOperation, Single, CudaSparseMatrixDescriptor, CudaSparseHybMat, CudaSparseSolveAnalysisInfo, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

Solution of triangular linear system op(A) * y = alpha * x, where A is a sparse matrix in HYB storage format, x and y are dense vectors.

Declaration

public void HybsvSolve(cusparseOperation transA, float alpha, CudaSparseMatrixDescriptor descrA, CudaSparseHybMat hybA, CudaSparseSolveAnalysisInfo info, CudaDeviceVariable<float> x, CudaDeviceVariable<float> y)

Parameters

Type	Name	Description
cusparseOperation	transA	the operation op(A) (currently only op(A) = A is supported).
System.Single	alpha	scalar used for multiplication.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_TRIANGULAR and diagonal type CUSPARSE_DIAG_TYPE_NON_UNIT.
CudaSparseHybMat	hybA	the matrix A in HYB storage format.
CudaSparseSolveAnalysisInfo	info	structure filled with information collected during the analysis phase (that should be passed to the solve phase unchanged).
CudaDeviceVariable<System.Single>	x	right-hand-side vector of size m.
CudaDeviceVariable<System.Single>	y	solution vector of size m.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Int32>, ref Int32)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, ref int nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
System.Int32	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Int32>, ref Int32)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, ref int nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
System.Int32	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Int32>, ref Int32)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, ref int nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Double>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
System.Int32	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz(cusparseDirection, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Int32>, ref Int32)

This routine finds the total number of non-zero elements and the number of non-zero elements per row or column in the dense matrix A.

Declaration

public void Nnz(cusparseDirection dirA, int m, int n, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<int> nnzPerRowCol, ref int nnzTotalDevHostPtr)

Parameters

Type	Name	Description
cusparseDirection	dirA	direction that specifies whether to count non-zero elements by CUSPARSE_DIRECTION_ROW or CUSPARSE_DIRECTION_COLUMN.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaSparseMatrixDescriptor	descrA	the descriptor of matrix A. The supported matrix type is CUSPARSE_MATRIX_TYPE_GENERAL. Also, the supported index bases are CUSPARSE_INDEX_BASE_ZERO and CUSPARSE_INDEX_BASE_ONE.
CudaDeviceVariable<System.Single>	A	array of dimensions (lda, n).
System.Int32	lda	leading dimension of dense array A.
CudaDeviceVariable<System.Int32>	nnzPerRowCol	Output: array of size m or n containing the number of non-zero elements per row or column, respectively.
System.Int32	nnzTotalDevHostPtr	Output: total number of non-zero elements in device or host memory.

Nnz_compress(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuDoubleComplex)

This routine finds the total number of non-zero elements and the number of non-zero elements per row in a noncompressed csr matrix A.

Declaration

public void Nnz_compress(int m, CudaSparseMatrixDescriptor descr, CudaDeviceVariable<cuDoubleComplex> values, CudaDeviceVariable<int> rowPtr, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<int> nnzTotal, cuDoubleComplex tol)

Parameters

Type	Name	Description
System.Int32	m
CudaSparseMatrixDescriptor	descr
CudaDeviceVariable<cuDoubleComplex>	values
CudaDeviceVariable<System.Int32>	rowPtr
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<System.Int32>	nnzTotal
cuDoubleComplex	tol

Nnz_compress(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, cuFloatComplex)

This routine finds the total number of non-zero elements and the number of non-zero elements per row in a noncompressed csr matrix A.

Declaration

public void Nnz_compress(int m, CudaSparseMatrixDescriptor descr, CudaDeviceVariable<cuFloatComplex> values, CudaDeviceVariable<int> rowPtr, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<int> nnzTotal, cuFloatComplex tol)

Parameters

Type	Name	Description
System.Int32	m
CudaSparseMatrixDescriptor	descr
CudaDeviceVariable<cuFloatComplex>	values
CudaDeviceVariable<System.Int32>	rowPtr
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<System.Int32>	nnzTotal
cuFloatComplex	tol

Nnz_compress(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double)

This routine finds the total number of non-zero elements and the number of non-zero elements per row in a noncompressed csr matrix A.

Declaration

public void Nnz_compress(int m, CudaSparseMatrixDescriptor descr, CudaDeviceVariable<double> values, CudaDeviceVariable<int> rowPtr, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<int> nnzTotal, double tol)

Parameters

Type	Name	Description
System.Int32	m
CudaSparseMatrixDescriptor	descr
CudaDeviceVariable<System.Double>	values
CudaDeviceVariable<System.Int32>	rowPtr
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<System.Int32>	nnzTotal
System.Double	tol

Nnz_compress(Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single)

This routine finds the total number of non-zero elements and the number of non-zero elements per row in a noncompressed csr matrix A.

Declaration

public void Nnz_compress(int m, CudaSparseMatrixDescriptor descr, CudaDeviceVariable<float> values, CudaDeviceVariable<int> rowPtr, CudaDeviceVariable<int> nnzPerRow, CudaDeviceVariable<int> nnzTotal, float tol)

Parameters

Type	Name	Description
System.Int32	m
CudaSparseMatrixDescriptor	descr
CudaDeviceVariable<System.Single>	values
CudaDeviceVariable<System.Int32>	rowPtr
CudaDeviceVariable<System.Int32>	nnzPerRow
CudaDeviceVariable<System.Int32>	nnzTotal
System.Single	tol

PruneCsr2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, half, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csr(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, half threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
half	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<half>, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csr(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<half> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<half>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csr(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Double>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csr(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, double threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Double	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csr(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Single>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csr(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csr(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csrBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, half, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune sparse matrix with CSR format to another sparse matrix with CSR format

Declaration

public SizeT PruneCsr2csrBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, half threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
half	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneCsr2csrBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<half>, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune sparse matrix with CSR format to another sparse matrix with CSR format

Declaration

public SizeT PruneCsr2csrBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<half> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<half>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneCsr2csrBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune sparse matrix with CSR format to another sparse matrix with CSR format

Declaration

public SizeT PruneCsr2csrBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Double>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneCsr2csrBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune sparse matrix with CSR format to another sparse matrix with CSR format

Declaration

public SizeT PruneCsr2csrBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, double threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Double	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneCsr2csrBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune sparse matrix with CSR format to another sparse matrix with CSR format

Declaration

public SizeT PruneCsr2csrBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Single>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneCsr2csrBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune sparse matrix with CSR format to another sparse matrix with CSR format

Declaration

public SizeT PruneCsr2csrBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneCsr2csrByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csrByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csrByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csrByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csrByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public void PruneCsr2csrByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

PruneCsr2csrByPercentageBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo)

Description: prune sparse matrix to a sparse matrix with CSR format by percentage

Declaration

public SizeT PruneCsr2csrByPercentageBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info

Returns

Type	Description
SizeT

PruneCsr2csrByPercentageBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo)

Description: prune sparse matrix to a sparse matrix with CSR format by percentage

Declaration

public SizeT PruneCsr2csrByPercentageBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info

Returns

Type	Description
SizeT

PruneCsr2csrByPercentageBufferSizeExt(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo)

Description: prune sparse matrix to a sparse matrix with CSR format by percentage

Declaration

public SizeT PruneCsr2csrByPercentageBufferSizeExt(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info

Returns

Type	Description
SizeT

PruneCsr2csrNnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, half, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public int PruneCsr2csrNnz(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, half threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
half	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneCsr2csrNnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<half>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneCsr2csrNnz(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<half> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<half>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneCsr2csrNnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneCsr2csrNnz(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<double> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Double>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneCsr2csrNnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public int PruneCsr2csrNnz(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, double threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Double	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneCsr2csrNnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneCsr2csrNnz(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, CudaDeviceVariable<float> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
CudaDeviceVariable<System.Single>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneCsr2csrNnz(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public int PruneCsr2csrNnz(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneCsr2csrNnzByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public int PruneCsr2csrNnzByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneCsr2csrNnzByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneCsr2csrNnzByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<half> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<half>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneCsr2csrNnzByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public int PruneCsr2csrNnzByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneCsr2csrNnzByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneCsr2csrNnzByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<double> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Double>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneCsr2csrNnzByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public int PruneCsr2csrNnzByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneCsr2csrNnzByPercentage(Int32, Int32, Int32, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneCsr2csrNnzByPercentage(int m, int n, int nnzA, CudaSparseMatrixDescriptor descrA, CudaDeviceVariable<float> csrValA, CudaDeviceVariable<int> csrRowPtrA, CudaDeviceVariable<int> csrColIndA, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
System.Int32	nnzA
CudaSparseMatrixDescriptor	descrA
CudaDeviceVariable<System.Single>	csrValA
CudaDeviceVariable<System.Int32>	csrRowPtrA
CudaDeviceVariable<System.Int32>	csrColIndA
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneDense2csr(Int32, Int32, CudaDeviceVariable<half>, Int32, half, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csr(int m, int n, CudaDeviceVariable<half> A, int lda, half threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
half	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csr(Int32, Int32, CudaDeviceVariable<half>, Int32, CudaDeviceVariable<half>, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csr(int m, int n, CudaDeviceVariable<half> A, int lda, CudaDeviceVariable<half> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
CudaDeviceVariable<half>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csr(Int32, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csr(int m, int n, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
CudaDeviceVariable<System.Double>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csr(Int32, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csr(int m, int n, CudaDeviceVariable<double> A, int lda, double threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
System.Double	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csr(Int32, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csr(int m, int n, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
CudaDeviceVariable<System.Single>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csr(Int32, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csr(int m, int n, CudaDeviceVariable<float> A, int lda, float threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
System.Single	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csrBufferSize(Int32, Int32, CudaDeviceVariable<half>, Int32, half, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune dense matrix to a sparse matrix with CSR format

Declaration

public SizeT PruneDense2csrBufferSize(int m, int n, CudaDeviceVariable<half> A, int lda, half threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
half	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneDense2csrBufferSize(Int32, Int32, CudaDeviceVariable<half>, Int32, CudaDeviceVariable<half>, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune dense matrix to a sparse matrix with CSR format

Declaration

public SizeT PruneDense2csrBufferSize(int m, int n, CudaDeviceVariable<half> A, int lda, CudaDeviceVariable<half> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
CudaDeviceVariable<half>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneDense2csrBufferSize(Int32, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune dense matrix to a sparse matrix with CSR format

Declaration

public SizeT PruneDense2csrBufferSize(int m, int n, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
CudaDeviceVariable<System.Double>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneDense2csrBufferSize(Int32, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune dense matrix to a sparse matrix with CSR format

Declaration

public SizeT PruneDense2csrBufferSize(int m, int n, CudaDeviceVariable<double> A, int lda, double threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
System.Double	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneDense2csrBufferSize(Int32, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune dense matrix to a sparse matrix with CSR format

Declaration

public SizeT PruneDense2csrBufferSize(int m, int n, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
CudaDeviceVariable<System.Single>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneDense2csrBufferSize(Int32, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>)

Description: prune dense matrix to a sparse matrix with CSR format

Declaration

public SizeT PruneDense2csrBufferSize(int m, int n, CudaDeviceVariable<float> A, int lda, float threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
System.Single	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC

Returns

Type	Description
SizeT

PruneDense2csrByPercentage(Int32, Int32, CudaDeviceVariable<half>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csrByPercentage(int m, int n, CudaDeviceVariable<half> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csrByPercentage(Int32, Int32, CudaDeviceVariable<Double>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csrByPercentage(int m, int n, CudaDeviceVariable<double> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csrByPercentage(Int32, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public void PruneDense2csrByPercentage(int m, int n, CudaDeviceVariable<float> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

PruneDense2csrByPercentageBufferSizeExt(Int32, Int32, CudaDeviceVariable<half>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<half>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo)

Description: prune dense matrix to a sparse matrix with CSR format by percentage

Declaration

public SizeT PruneDense2csrByPercentageBufferSizeExt(int m, int n, CudaDeviceVariable<half> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<half> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<half>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info

Returns

Type	Description
SizeT

PruneDense2csrByPercentageBufferSizeExt(Int32, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo)

Description: prune dense matrix to a sparse matrix with CSR format by percentage

Declaration

public SizeT PruneDense2csrByPercentageBufferSizeExt(int m, int n, CudaDeviceVariable<float> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<float> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Single>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info

Returns

Type	Description
SizeT

PruneDense2csrByPercentageBufferSizeExtt(Int32, Int32, CudaDeviceVariable<Double>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, CudaSparsePruneInfo)

Description: prune dense matrix to a sparse matrix with CSR format by percentage

Declaration

public SizeT PruneDense2csrByPercentageBufferSizeExtt(int m, int n, CudaDeviceVariable<double> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<double> csrValC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<int> csrColIndC, CudaSparsePruneInfo info)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Double>	csrValC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Int32>	csrColIndC
CudaSparsePruneInfo	info

Returns

Type	Description
SizeT

PruneDense2csrNnz(Int32, Int32, CudaDeviceVariable<half>, Int32, half, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public int PruneDense2csrNnz(int m, int n, CudaDeviceVariable<half> A, int lda, half threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
half	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneDense2csrNnz(Int32, Int32, CudaDeviceVariable<half>, Int32, CudaDeviceVariable<half>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneDense2csrNnz(int m, int n, CudaDeviceVariable<half> A, int lda, CudaDeviceVariable<half> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
CudaDeviceVariable<half>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneDense2csrNnz(Int32, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneDense2csrNnz(int m, int n, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
CudaDeviceVariable<System.Double>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneDense2csrNnz(Int32, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public int PruneDense2csrNnz(int m, int n, CudaDeviceVariable<double> A, int lda, double threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
System.Double	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneDense2csrNnz(Int32, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneDense2csrNnz(int m, int n, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
CudaDeviceVariable<System.Single>	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneDense2csrNnz(Int32, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaDeviceVariable<Byte>)

Declaration

public int PruneDense2csrNnz(int m, int n, CudaDeviceVariable<float> A, int lda, float threshold, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
System.Single	threshold
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneDense2csrNnzByPercentage(Int32, Int32, CudaDeviceVariable<half>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public int PruneDense2csrNnzByPercentage(int m, int n, CudaDeviceVariable<half> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneDense2csrNnzByPercentage(Int32, Int32, CudaDeviceVariable<half>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneDense2csrNnzByPercentage(int m, int n, CudaDeviceVariable<half> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<half>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneDense2csrNnzByPercentage(Int32, Int32, CudaDeviceVariable<Double>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public int PruneDense2csrNnzByPercentage(int m, int n, CudaDeviceVariable<double> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneDense2csrNnzByPercentage(Int32, Int32, CudaDeviceVariable<Double>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneDense2csrNnzByPercentage(int m, int n, CudaDeviceVariable<double> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Double>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

PruneDense2csrNnzByPercentage(Int32, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>)

Declaration

public int PruneDense2csrNnzByPercentage(int m, int n, CudaDeviceVariable<float> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer

Returns

Type	Description
System.Int32

PruneDense2csrNnzByPercentage(Int32, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaSparseMatrixDescriptor, CudaDeviceVariable<Int32>, CudaSparsePruneInfo, CudaDeviceVariable<Byte>, CudaDeviceVariable<Int32>)

Declaration

public void PruneDense2csrNnzByPercentage(int m, int n, CudaDeviceVariable<float> A, int lda, float percentage, CudaSparseMatrixDescriptor descrC, CudaDeviceVariable<int> csrRowPtrC, CudaSparsePruneInfo info, CudaDeviceVariable<byte> pBuffer, CudaDeviceVariable<int> nnzTotalDevHostPtr)

Parameters

Type	Name	Description
System.Int32	m
System.Int32	n
CudaDeviceVariable<System.Single>	A
System.Int32	lda
System.Single	percentage
CudaSparseMatrixDescriptor	descrC
CudaDeviceVariable<System.Int32>	csrRowPtrC
CudaSparsePruneInfo	info
CudaDeviceVariable<System.Byte>	pBuffer
CudaDeviceVariable<System.Int32>	nnzTotalDevHostPtr

Roti(CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, cusparseIndexBase)

Givens rotation, where c and s are cosine and sine, x and y are sparse and dense vectors, respectively.

Declaration

public void Roti(CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> y, CudaDeviceVariable<double> c, CudaDeviceVariable<double> s, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	y	vector in dense format.
CudaDeviceVariable<System.Double>	c	cosine element of the rotation matrix.
CudaDeviceVariable<System.Double>	s	sine element of the rotation matrix.
cusparseIndexBase	idxBase	Index base.

Roti(CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, Double, Double, cusparseIndexBase)

Givens rotation, where c and s are cosine and sine, x and y are sparse and dense vectors, respectively.

Declaration

public void Roti(CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> y, double c, double s, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	y	vector in dense format.
System.Double	c	cosine element of the rotation matrix.
System.Double	s	sine element of the rotation matrix.
cusparseIndexBase	idxBase	Index base.

Roti(CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, cusparseIndexBase)

Givens rotation, where c and s are cosine and sine, x and y are sparse and dense vectors, respectively.

Declaration

public void Roti(CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> y, CudaDeviceVariable<float> c, CudaDeviceVariable<float> s, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	y	vector in dense format.
CudaDeviceVariable<System.Single>	c	cosine element of the rotation matrix.
CudaDeviceVariable<System.Single>	s	sine element of the rotation matrix.
cusparseIndexBase	idxBase	Index base.

Roti(CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, Single, Single, cusparseIndexBase)

Givens rotation, where c and s are cosine and sine, x and y are sparse and dense vectors, respectively.

Declaration

public void Roti(CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> y, float c, float s, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	y	vector in dense format.
System.Single	c	cosine element of the rotation matrix.
System.Single	s	sine element of the rotation matrix.
cusparseIndexBase	idxBase	Index base.

Sctr(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuDoubleComplex>, cusparseIndexBase)

Scatter of elements of the sparse vector x into dense vector y.

Declaration

public void Sctr(CudaDeviceVariable<cuDoubleComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuDoubleComplex> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuDoubleComplex>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Sctr(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Int32>, CudaDeviceVariable<cuFloatComplex>, cusparseIndexBase)

Scatter of elements of the sparse vector x into dense vector y.

Declaration

public void Sctr(CudaDeviceVariable<cuFloatComplex> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<cuFloatComplex> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<cuFloatComplex>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Sctr(CudaDeviceVariable<Double>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Double>, cusparseIndexBase)

Scatter of elements of the sparse vector x into dense vector y.

Declaration

public void Sctr(CudaDeviceVariable<double> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<double> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Double>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

Sctr(CudaDeviceVariable<Single>, CudaDeviceVariable<Int32>, CudaDeviceVariable<Single>, cusparseIndexBase)

Scatter of elements of the sparse vector x into dense vector y.

Declaration

public void Sctr(CudaDeviceVariable<float> xVal, CudaDeviceVariable<int> xInd, CudaDeviceVariable<float> y, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	xVal	vector with nnz non-zero values of vector x.
CudaDeviceVariable<System.Int32>	xInd	integer vector with nnz indices of the non-zero values of vector x. Length of xInd gives the number nzz passed to CUSPARSE.
CudaDeviceVariable<System.Single>	y	vector in dense format.
cusparseIndexBase	idxBase	Index base.

SetPointerMode(cusparsePointerMode)

Sets the pointer mode for scalar values (host or device pointer)

Declaration

public void SetPointerMode(cusparsePointerMode pointerMode)

Parameters

Type	Name	Description
cusparsePointerMode	pointerMode

SetStream(CUstream)

Sets the cuda stream to use

Declaration

public void SetStream(CUstream stream)

Parameters

Type	Name	Description
CUstream	stream	A valid CUDA stream created with cudaStreamCreate() (or 0 for the default stream)

Xbsrsm2ZeroPivot(CudaSparseBsrsm2Info, CudaDeviceVariable<Int32>)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) is either a structural zero or a numerical zero (singular block). Otherwise position=-1.

The position can be 0-base or 1-base, the same as the matrix. Function cusparseXbsrsm2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Xbsrsm2ZeroPivot(CudaSparseBsrsm2Info info, CudaDeviceVariable<int> position)

Parameters

Type	Name	Description
CudaSparseBsrsm2Info	info	info contains a structural zero or a numerical zero if the user already called bsrsm2_analysis() or bsrsm2_solve().
CudaDeviceVariable<System.Int32>	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Xbsrsm2ZeroPivot(CudaSparseBsrsm2Info, ref Int32)

If the returned error code is CUSPARSE_STATUS_ZERO_PIVOT, position=j means A(j,j) is either a structural zero or a numerical zero (singular block). Otherwise position=-1.

The position can be 0-base or 1-base, the same as the matrix. Function cusparseXbsrsm2_zeroPivot() is a blocking call. It calls cudaDeviceSynchronize() to make sure all previous kernels are done.

The position can be in the host memory or device memory. The user can set the proper mode with cusparseSetPointerMode().

Declaration

public bool Xbsrsm2ZeroPivot(CudaSparseBsrsm2Info info, ref int position)

Parameters

Type	Name	Description
CudaSparseBsrsm2Info	info	info contains a structural zero or a numerical zero if the user already called bsrsm2_analysis() or bsrsm2_solve().
System.Int32	position	if no structural or numerical zero, position is -1; otherwise, if A(j,j) is missing or U(j,j) is zero, position=j.

Returns

Type	Description
System.Boolean	If true, position=j means A(j,j) has either a structural zero or a numerical zero; otherwise, position=-1.

Xcoo2csr(CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Int32>, cusparseIndexBase)

This routine compresses the indecis of rows or columns. It can be interpreted as a conversion from COO to CSR sparse storage format.

Declaration

public void Xcoo2csr(CudaDeviceVariable<int> cooRowInd, int m, CudaDeviceVariable<int> csrRowPtr, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Int32>	cooRowInd	integer array of nnz uncompressed row indices. Length of cooRowInd gives the number nzz passed to CUSPARSE.
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<System.Int32>	csrRowPtr	Output: integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
cusparseIndexBase	idxBase	Index base.

Xcsr2coo(CudaDeviceVariable<Int32>, Int32, CudaDeviceVariable<Int32>, cusparseIndexBase)

This routine uncompresses the indecis of rows or columns. It can be interpreted as a conversion from CSR to COO sparse storage format.

Declaration

public void Xcsr2coo(CudaDeviceVariable<int> csrRowPtr, int m, CudaDeviceVariable<int> cooRowInd, cusparseIndexBase idxBase)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Int32>	csrRowPtr	Output: integer array of m + 1 elements that contains the start of every row and the end of the last row plus one.
System.Int32	m	number of rows of matrix A.
CudaDeviceVariable<System.Int32>	cooRowInd	integer array of nnz uncompressed row indices. Length of cooRowInd gives the number nzz passed to CUSPARSE.
cusparseIndexBase	idxBase	Index base.

Implements

System.IDisposable