Class CudaSolveNativeMethods.Dense

The cuSolverDN library was designed to solve dense linear systems of the form Ax=B

Inheritance

System.Object

CudaSolveNativeMethods.Dense

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.CudaSolve

Assembly: CudaSolve.dll

Syntax

public static class Dense

Methods

cusolverDnCgebrd(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnCgebrd(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr D, CUdeviceptr E, CUdeviceptr TAUQ, CUdeviceptr TAUP, CUdeviceptr Work, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	D	array of dimension min(m,n). The diagonal elements of the bidiagonal matrix B: D(i) = A(i,i).
CUdeviceptr	E	array of dimension min(m,n). The off-diagonal elements of the bidiagonal matrix B: if m>=n, E(i) = A(i,i+1) for i = 1,2,...,n-1; if m<n, E(i) = A(i+1,i) for i = 1,2,...,m-1.
CUdeviceptr	TAUQ	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix Q.
CUdeviceptr	TAUP	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix P.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCgebrd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnCgebrd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnCgeqrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnCgeqrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr TAU, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	TAU	array of dimension at least min(m,n).
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of working array Workspace.
CUdeviceptr	devInfo	if info = 0, the LU factorization is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCgeqrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnCgeqrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnCgesvd(cusolverDnHandle, Char, Char, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnCgesvd(cusolverDnHandle handle, char jobu, char jobvt, int m, int n, CUdeviceptr A, int lda, CUdeviceptr S, CUdeviceptr U, int ldu, CUdeviceptr VT, int ldvt, CUdeviceptr Work, int Lwork, CUdeviceptr rwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Char	jobu	specifies options for computing all or part of the matrix U: = 'A': all m columns of U are returned in array U: = 'S': the first min(m,n) columns of U (the left singular vectors) are returned in the array U; = 'O': the first min(m,n) columns of U (the left singular vectors) are overwritten on the array A; = 'N': no columns of U (no left singular vectors) are computed.
System.Char	jobvt	specifies options for computing all or part of the matrix VT: = 'A': all N rows of VT are returned in the array VT; = 'S': the first min(m,n) rows of VT (the right singular vectors) are returned in the array VT; = 'O': the first min(m,n) rows of VT (the right singular vectors) are overwritten on the array A; = 'N': no rows of V**T (no right singular vectors) are computed.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). On exit, the contents of A are destroyed.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	S	array of dimension min(m,n). The singular values of A, sorted so that S(i) = S(i+1).
CUdeviceptr	U	array of dimension ldu * m with ldu is not less than max(1,m). U contains the m×m unitary matrix U.
System.Int32	ldu	leading dimension of two-dimensional array used to store matrix U.
CUdeviceptr	VT	array of dimension ldvt * n with ldvt is not less than max(1,n). VT contains the n×n unitary matrix V**T.
System.Int32	ldvt	leading dimension of two-dimensional array used to store matrix Vt.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.
CUdeviceptr	rwork
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the ith parameter is wrong. if devInfo > 0, devInfo indicates how many superdiagonals of an intermediate bidiagonal form B did not converge to zero.

Returns

Type	Description
cusolverStatus

cusolverDnCgesvd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnCgesvd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnCgetrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnCgetrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, CUdeviceptr devIpiv, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
CUdeviceptr	devIpiv	array of size at least min(m,n), containing pivot indices.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the U(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnCgetrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnCgetrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnCgetrs(cusolverDnHandle, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function solves a linear system of multiple right-hand sides op(A)*X=B.

Declaration

public static cusolverStatus cusolverDnCgetrs(cusolverDnHandle handle, Operation trans, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr devIpiv, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of right-hand sides.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	devIpiv	array of size at least n, containing pivot indices.
CUdeviceptr	B	array of dimension ldb * nrhs with ldb is not less than max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCheevd(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnCheevd(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCheevd_bufferSize(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnCheevd_bufferSize(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work.

Returns

Type	Description
cusolverStatus

cusolverDnChegvd(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix-pair (A,B).

Declaration

public static cusolverStatus cusolverDnChegvd(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnChegvd_bufferSize(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

The helper functions below can calculate the sizes needed for pre-allocated buffer.

Declaration

public static cusolverStatus cusolverDnChegvd_bufferSize(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work

Returns

Type	Description
cusolverStatus

cusolverDnChetrd(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

Declaration

public static cusolverStatus cusolverDnChetrd(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr d, CUdeviceptr e, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle
FillMode	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	d
CUdeviceptr	e
CUdeviceptr	tau
CUdeviceptr	work
System.Int32	lwork
CUdeviceptr	info

Returns

Type	Description
cusolverStatus

cusolverDnChetrd_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, ref Int32)

Declaration

public static cusolverStatus cusolverDnChetrd_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr d, CUdeviceptr e, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle
FillMode	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	d
CUdeviceptr	e
CUdeviceptr	tau
System.Int32	lwork

Returns

Type	Description
cusolverStatus

cusolverDnClaswp(cusolverDnHandle, Int32, CUdeviceptr, Int32, Int32, Int32, CUdeviceptr, Int32)

Declaration

public static cusolverStatus cusolverDnClaswp(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, int k1, int k2, CUdeviceptr devIpiv, int incx)

Parameters

Type	Name	Description
cusolverDnHandle	handle
System.Int32	n
CUdeviceptr	A
System.Int32	lda
System.Int32	k1
System.Int32	k2
CUdeviceptr	devIpiv
System.Int32	incx

Returns

Type	Description
cusolverStatus

cusolverDnCpotrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnCpotrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of Workspace
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the leading minor of order i is not positive definite.

Returns

Type	Description
cusolverStatus

cusolverDnCpotrf_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnCpotrf_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnCpotrs(cusolverDnHandle, FillMode, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function solves a system of linear equations A*X=B where A is a n×n Hermitian matrix, only lower or upper part is meaningful.

Declaration

public static cusolverStatus cusolverDnCpotrs(cusolverDnHandle handle, FillMode uplo, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of columns of matrix X and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n). A is either lower cholesky factor L or upper Cholesky factor U.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	B	array of dimension ldb * nrhs. ldb is not less than max(1,n). As an input, B is right hand side matrix. As an output, B is the solution matrix.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCreate(ref cusolverDnHandle)

This function initializes the cuSolverDN library and creates a handle on the cuSolverDN context. It must be called before any other cuSolverDN API function is invoked. It allocates hardware resources necessary for accessing the GPU

Declaration

public static cusolverStatus cusolverDnCreate(ref cusolverDnHandle handle)

Parameters

Type	Name	Description
cusolverDnHandle	handle	the pointer to the handle to the cuSolverDN context.

Returns

Type	Description
cusolverStatus

cusolverDnCsytrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnCsytrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr ipiv, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	ipiv	array of size at least n, containing pivot indices.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working space work.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the D(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnCsytrf_bufferSize(cusolverDnHandle, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnCsytrf_bufferSize(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working space work.

Returns

Type	Description
cusolverStatus

cusolverDnCungbr(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnCungbr(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCungbr_bufferSize(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnCungbr_bufferSize(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnCungqr(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnCungqr(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, rray of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCungqr_bufferSize(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnCungqr_bufferSize(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnCungtr(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnCungtr(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgtr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCungtr_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnCungtr_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnCunmqr(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnCunmqr(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n). The vector tau is from geqrf, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	devInfo	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCunmqr_bufferSize(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnCunmqr_bufferSize(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * k with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n).
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnCunmtr(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnCunmtr(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnCunmtr_bufferSize(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnCunmtr_bufferSize(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnDestroy(cusolverDnHandle)

This function releases CPU-side resources used by the cuSolverDN library.

Declaration

public static cusolverStatus cusolverDnDestroy(cusolverDnHandle handle)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.

Returns

Type	Description
cusolverStatus

cusolverDnDgebrd(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnDgebrd(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr D, CUdeviceptr E, CUdeviceptr TAUQ, CUdeviceptr TAUP, CUdeviceptr Work, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	D	array of dimension min(m,n). The diagonal elements of the bidiagonal matrix B: D(i) = A(i,i).
CUdeviceptr	E	array of dimension min(m,n). The off-diagonal elements of the bidiagonal matrix B: if m>=n, E(i) = A(i,i+1) for i = 1,2,...,n-1; if m<n, E(i) = A(i+1,i) for i = 1,2,...,m-1.
CUdeviceptr	TAUQ	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix Q.
CUdeviceptr	TAUP	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix P.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDgebrd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnDgebrd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnDgeqrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnDgeqrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr TAU, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	TAU	array of dimension at least min(m,n).
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of working array Workspace.
CUdeviceptr	devInfo	if info = 0, the LU factorization is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDgeqrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnDgeqrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnDgesvd(cusolverDnHandle, Char, Char, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnDgesvd(cusolverDnHandle handle, char jobu, char jobvt, int m, int n, CUdeviceptr A, int lda, CUdeviceptr S, CUdeviceptr U, int ldu, CUdeviceptr VT, int ldvt, CUdeviceptr Work, int Lwork, CUdeviceptr rwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Char	jobu	specifies options for computing all or part of the matrix U: = 'A': all m columns of U are returned in array U: = 'S': the first min(m,n) columns of U (the left singular vectors) are returned in the array U; = 'O': the first min(m,n) columns of U (the left singular vectors) are overwritten on the array A; = 'N': no columns of U (no left singular vectors) are computed.
System.Char	jobvt	specifies options for computing all or part of the matrix VT: = 'A': all N rows of VT are returned in the array VT; = 'S': the first min(m,n) rows of VT (the right singular vectors) are returned in the array VT; = 'O': the first min(m,n) rows of VT (the right singular vectors) are overwritten on the array A; = 'N': no rows of V**T (no right singular vectors) are computed.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). On exit, the contents of A are destroyed.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	S	array of dimension min(m,n). The singular values of A, sorted so that S(i) = S(i+1).
CUdeviceptr	U	array of dimension ldu * m with ldu is not less than max(1,m). U contains the m×m unitary matrix U.
System.Int32	ldu	leading dimension of two-dimensional array used to store matrix U.
CUdeviceptr	VT	array of dimension ldvt * n with ldvt is not less than max(1,n). VT contains the n×n unitary matrix V**T.
System.Int32	ldvt	leading dimension of two-dimensional array used to store matrix Vt.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.
CUdeviceptr	rwork
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the ith parameter is wrong. if devInfo > 0, devInfo indicates how many superdiagonals of an intermediate bidiagonal form B did not converge to zero.

Returns

Type	Description
cusolverStatus

cusolverDnDgesvd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnDgesvd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnDgetrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnDgetrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, CUdeviceptr devIpiv, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
CUdeviceptr	devIpiv	array of size at least min(m,n), containing pivot indices.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the U(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnDgetrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnDgetrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnDgetrs(cusolverDnHandle, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function solves a linear system of multiple right-hand sides op(A)*X=B.

Declaration

public static cusolverStatus cusolverDnDgetrs(cusolverDnHandle handle, Operation trans, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr devIpiv, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of right-hand sides.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	devIpiv	array of size at least n, containing pivot indices.
CUdeviceptr	B	array of dimension ldb * nrhs with ldb is not less than max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDlaswp(cusolverDnHandle, Int32, CUdeviceptr, Int32, Int32, Int32, CUdeviceptr, Int32)

Declaration

public static cusolverStatus cusolverDnDlaswp(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, int k1, int k2, CUdeviceptr devIpiv, int incx)

Parameters

Type	Name	Description
cusolverDnHandle	handle
System.Int32	n
CUdeviceptr	A
System.Int32	lda
System.Int32	k1
System.Int32	k2
CUdeviceptr	devIpiv
System.Int32	incx

Returns

Type	Description
cusolverStatus

cusolverDnDorgbr(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnDorgbr(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDorgbr_bufferSize(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnDorgbr_bufferSize(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnDorgqr(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnDorgqr(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, rray of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDorgqr_bufferSize(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnDorgqr_bufferSize(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnDorgtr(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnDorgtr(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgtr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDorgtr_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnDorgtr_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnDormqr(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnDormqr(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n). The vector tau is from geqrf, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	devInfo	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDormqr_bufferSize(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnDormqr_bufferSize(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * k with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n).
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnDormtr(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnDormtr(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDormtr_bufferSize(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnDormtr_bufferSize(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnDpotrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnDpotrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of Workspace
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the leading minor of order i is not positive definite.

Returns

Type	Description
cusolverStatus

cusolverDnDpotrf_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnDpotrf_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnDpotrs(cusolverDnHandle, FillMode, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function solves a system of linear equations A*X=B where A is a n×n Hermitian matrix, only lower or upper part is meaningful.

Declaration

public static cusolverStatus cusolverDnDpotrs(cusolverDnHandle handle, FillMode uplo, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of columns of matrix X and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n). A is either lower cholesky factor L or upper Cholesky factor U.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	B	array of dimension ldb * nrhs. ldb is not less than max(1,n). As an input, B is right hand side matrix. As an output, B is the solution matrix.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDsyevd(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnDsyevd(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDsyevd_bufferSize(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnDsyevd_bufferSize(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work.

Returns

Type	Description
cusolverStatus

cusolverDnDsygvd(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix-pair (A,B).

Declaration

public static cusolverStatus cusolverDnDsygvd(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnDsygvd_bufferSize(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

The helper functions below can calculate the sizes needed for pre-allocated buffer.

Declaration

public static cusolverStatus cusolverDnDsygvd_bufferSize(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work

Returns

Type	Description
cusolverStatus

cusolverDnDsytrd(cusolverDnHandle, Char, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

Declaration

public static cusolverStatus cusolverDnDsytrd(cusolverDnHandle handle, char uplo, int n, CUdeviceptr A, int lda, CUdeviceptr D, CUdeviceptr E, CUdeviceptr tau, CUdeviceptr Work, int Lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle
System.Char	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	D
CUdeviceptr	E
CUdeviceptr	tau
CUdeviceptr	Work
System.Int32	Lwork
CUdeviceptr	info

Returns

Type	Description
cusolverStatus

cusolverDnDsytrd_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, ref Int32)

Declaration

public static cusolverStatus cusolverDnDsytrd_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr d, CUdeviceptr e, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle
FillMode	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	d
CUdeviceptr	e
CUdeviceptr	tau
System.Int32	lwork

Returns

Type	Description
cusolverStatus

cusolverDnDsytrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnDsytrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr ipiv, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	ipiv	array of size at least n, containing pivot indices.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working space work.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the D(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnDsytrf_bufferSize(cusolverDnHandle, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnDsytrf_bufferSize(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working space work.

Returns

Type	Description
cusolverStatus

cusolverDnGetStream(cusolverDnHandle, ref CUstream)

This function sets the stream to be used by the cuSolverDN library to execute its routines.

Declaration

public static cusolverStatus cusolverDnGetStream(cusolverDnHandle handle, ref CUstream sreamId)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
CUstream	sreamId	the stream to be used by the library.

Returns

Type	Description
cusolverStatus

cusolverDnSetStream(cusolverDnHandle, CUstream)

This function sets the stream to be used by the cuSolverDN library to execute its routines.

Declaration

public static cusolverStatus cusolverDnSetStream(cusolverDnHandle handle, CUstream streamId)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
CUstream	streamId	the stream to be used by the library.

Returns

Type	Description
cusolverStatus

cusolverDnSgebrd(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnSgebrd(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr D, CUdeviceptr E, CUdeviceptr TAUQ, CUdeviceptr TAUP, CUdeviceptr Work, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	D	array of dimension min(m,n). The diagonal elements of the bidiagonal matrix B: D(i) = A(i,i).
CUdeviceptr	E	array of dimension min(m,n). The off-diagonal elements of the bidiagonal matrix B: if m>=n, E(i) = A(i,i+1) for i = 1,2,...,n-1; if m<n, E(i) = A(i+1,i) for i = 1,2,...,m-1.
CUdeviceptr	TAUQ	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix Q.
CUdeviceptr	TAUP	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix P.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSgebrd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnSgebrd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnSgeqrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnSgeqrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr TAU, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	TAU	array of dimension at least min(m,n).
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of working array Workspace.
CUdeviceptr	devInfo	if info = 0, the LU factorization is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSgeqrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnSgeqrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnSgesvd(cusolverDnHandle, Char, Char, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnSgesvd(cusolverDnHandle handle, char jobu, char jobvt, int m, int n, CUdeviceptr A, int lda, CUdeviceptr S, CUdeviceptr U, int ldu, CUdeviceptr VT, int ldvt, CUdeviceptr Work, int Lwork, CUdeviceptr rwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Char	jobu	specifies options for computing all or part of the matrix U: = 'A': all m columns of U are returned in array U: = 'S': the first min(m,n) columns of U (the left singular vectors) are returned in the array U; = 'O': the first min(m,n) columns of U (the left singular vectors) are overwritten on the array A; = 'N': no columns of U (no left singular vectors) are computed.
System.Char	jobvt	specifies options for computing all or part of the matrix VT: = 'A': all N rows of VT are returned in the array VT; = 'S': the first min(m,n) rows of VT (the right singular vectors) are returned in the array VT; = 'O': the first min(m,n) rows of VT (the right singular vectors) are overwritten on the array A; = 'N': no rows of V**T (no right singular vectors) are computed.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). On exit, the contents of A are destroyed.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	S	array of dimension min(m,n). The singular values of A, sorted so that S(i) = S(i+1).
CUdeviceptr	U	array of dimension ldu * m with ldu is not less than max(1,m). U contains the m×m unitary matrix U.
System.Int32	ldu	leading dimension of two-dimensional array used to store matrix U.
CUdeviceptr	VT	array of dimension ldvt * n with ldvt is not less than max(1,n). VT contains the n×n unitary matrix V**T.
System.Int32	ldvt	leading dimension of two-dimensional array used to store matrix Vt.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.
CUdeviceptr	rwork
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the ith parameter is wrong. if devInfo > 0, devInfo indicates how many superdiagonals of an intermediate bidiagonal form B did not converge to zero.

Returns

Type	Description
cusolverStatus

cusolverDnSgesvd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnSgesvd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnSgetrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnSgetrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, CUdeviceptr devIpiv, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
CUdeviceptr	devIpiv	array of size at least min(m,n), containing pivot indices.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the U(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnSgetrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnSgetrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnSgetrs(cusolverDnHandle, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function solves a linear system of multiple right-hand sides op(A)*X=B.

Declaration

public static cusolverStatus cusolverDnSgetrs(cusolverDnHandle handle, Operation trans, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr devIpiv, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of right-hand sides.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	devIpiv	array of size at least n, containing pivot indices.
CUdeviceptr	B	array of dimension ldb * nrhs with ldb is not less than max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSlaswp(cusolverDnHandle, Int32, CUdeviceptr, Int32, Int32, Int32, CUdeviceptr, Int32)

Declaration

public static cusolverStatus cusolverDnSlaswp(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, int k1, int k2, CUdeviceptr devIpiv, int incx)

Parameters

Type	Name	Description
cusolverDnHandle	handle
System.Int32	n
CUdeviceptr	A
System.Int32	lda
System.Int32	k1
System.Int32	k2
CUdeviceptr	devIpiv
System.Int32	incx

Returns

Type	Description
cusolverStatus

cusolverDnSorgbr(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnSorgbr(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSorgbr_bufferSize(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnSorgbr_bufferSize(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnSorgqr(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnSorgqr(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, rray of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSorgqr_bufferSize(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnSorgqr_bufferSize(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnSorgtr(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnSorgtr(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgtr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSorgtr_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnSorgtr_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnSormqr(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnSormqr(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n). The vector tau is from geqrf, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	devInfo	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSormqr_bufferSize(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnSormqr_bufferSize(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * k with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n).
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnSormtr(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnSormtr(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSormtr_bufferSize(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnSormtr_bufferSize(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnSpotrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnSpotrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of Workspace
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the leading minor of order i is not positive definite.

Returns

Type	Description
cusolverStatus

cusolverDnSpotrf_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnSpotrf_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnSpotrs(cusolverDnHandle, FillMode, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function solves a system of linear equations A*X=B where A is a n×n Hermitian matrix, only lower or upper part is meaningful.

Declaration

public static cusolverStatus cusolverDnSpotrs(cusolverDnHandle handle, FillMode uplo, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of columns of matrix X and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n). A is either lower cholesky factor L or upper Cholesky factor U.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	B	array of dimension ldb * nrhs. ldb is not less than max(1,n). As an input, B is right hand side matrix. As an output, B is the solution matrix.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSsyevd(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnSsyevd(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSsyevd_bufferSize(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnSsyevd_bufferSize(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work.

Returns

Type	Description
cusolverStatus

cusolverDnSsygvd(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix-pair (A,B).

Declaration

public static cusolverStatus cusolverDnSsygvd(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnSsygvd_bufferSize(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

The helper functions below can calculate the sizes needed for pre-allocated buffer.

Declaration

public static cusolverStatus cusolverDnSsygvd_bufferSize(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work

Returns

Type	Description
cusolverStatus

cusolverDnSsytrd(cusolverDnHandle, Char, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

Declaration

public static cusolverStatus cusolverDnSsytrd(cusolverDnHandle handle, char uplo, int n, CUdeviceptr A, int lda, CUdeviceptr D, CUdeviceptr E, CUdeviceptr tau, CUdeviceptr Work, int Lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle
System.Char	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	D
CUdeviceptr	E
CUdeviceptr	tau
CUdeviceptr	Work
System.Int32	Lwork
CUdeviceptr	info

Returns

Type	Description
cusolverStatus

cusolverDnSsytrd_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, ref Int32)

Declaration

public static cusolverStatus cusolverDnSsytrd_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr d, CUdeviceptr e, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle
FillMode	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	d
CUdeviceptr	e
CUdeviceptr	tau
System.Int32	lwork

Returns

Type	Description
cusolverStatus

cusolverDnSsytrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnSsytrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr ipiv, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	ipiv	array of size at least n, containing pivot indices.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working space work.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the D(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnSsytrf_bufferSize(cusolverDnHandle, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnSsytrf_bufferSize(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working space work.

Returns

Type	Description
cusolverStatus

cusolverDnZgebrd(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnZgebrd(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr D, CUdeviceptr E, CUdeviceptr TAUQ, CUdeviceptr TAUP, CUdeviceptr Work, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	D	array of dimension min(m,n). The diagonal elements of the bidiagonal matrix B: D(i) = A(i,i).
CUdeviceptr	E	array of dimension min(m,n). The off-diagonal elements of the bidiagonal matrix B: if m>=n, E(i) = A(i,i+1) for i = 1,2,...,n-1; if m<n, E(i) = A(i+1,i) for i = 1,2,...,m-1.
CUdeviceptr	TAUQ	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix Q.
CUdeviceptr	TAUP	array of dimension min(m,n). The scalar factors of the elementary reflectors which represent the orthogonal matrix P.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZgebrd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function reduces a general real m×n matrix A to upper or lower bidiagonal form B by an orthogonal transformation: Q^HAP=B

Declaration

public static cusolverStatus cusolverDnZgebrd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gebrd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnZgeqrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnZgeqrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr TAU, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	TAU	array of dimension at least min(m,n).
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of working array Workspace.
CUdeviceptr	devInfo	if info = 0, the LU factorization is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZgeqrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnZgeqrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnZgesvd(cusolverDnHandle, Char, Char, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnZgesvd(cusolverDnHandle handle, char jobu, char jobvt, int m, int n, CUdeviceptr A, int lda, CUdeviceptr S, CUdeviceptr U, int ldu, CUdeviceptr VT, int ldvt, CUdeviceptr Work, int Lwork, CUdeviceptr rwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Char	jobu	specifies options for computing all or part of the matrix U: = 'A': all m columns of U are returned in array U: = 'S': the first min(m,n) columns of U (the left singular vectors) are returned in the array U; = 'O': the first min(m,n) columns of U (the left singular vectors) are overwritten on the array A; = 'N': no columns of U (no left singular vectors) are computed.
System.Char	jobvt	specifies options for computing all or part of the matrix VT: = 'A': all N rows of VT are returned in the array VT; = 'S': the first min(m,n) rows of VT (the right singular vectors) are returned in the array VT; = 'O': the first min(m,n) rows of VT (the right singular vectors) are overwritten on the array A; = 'N': no rows of V**T (no right singular vectors) are computed.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). On exit, the contents of A are destroyed.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	S	array of dimension min(m,n). The singular values of A, sorted so that S(i) = S(i+1).
CUdeviceptr	U	array of dimension ldu * m with ldu is not less than max(1,m). U contains the m×m unitary matrix U.
System.Int32	ldu	leading dimension of two-dimensional array used to store matrix U.
CUdeviceptr	VT	array of dimension ldvt * n with ldvt is not less than max(1,n). VT contains the n×n unitary matrix V**T.
System.Int32	ldvt	leading dimension of two-dimensional array used to store matrix Vt.
CUdeviceptr	Work	working space, array of size Lwork.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.
CUdeviceptr	rwork
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the ith parameter is wrong. if devInfo > 0, devInfo indicates how many superdiagonals of an intermediate bidiagonal form B did not converge to zero.

Returns

Type	Description
cusolverStatus

cusolverDnZgesvd_bufferSize(cusolverDnHandle, Int32, Int32, ref Int32)

This function computes the singular value decomposition (SVD) of a m×n matrix A and corresponding the left and/or right singular vectors.

Declaration

public static cusolverStatus cusolverDnZgesvd_bufferSize(cusolverDnHandle handle, int m, int n, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	Lwork	size of Work, returned by gesvd_bufferSize.

Returns

Type	Description
cusolverStatus

cusolverDnZgetrf(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnZgetrf(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, CUdeviceptr devIpiv, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
CUdeviceptr	devIpiv	array of size at least min(m,n), containing pivot indices.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the U(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnZgetrf_bufferSize(cusolverDnHandle, Int32, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the LU factorization of a m×n matrix PA=LU where A is a m×n matrix, P is a permutation matrix, L is a lower triangular matrix with unit diagonal, and U is an upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnZgetrf_bufferSize(cusolverDnHandle handle, int m, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnZgetrs(cusolverDnHandle, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function solves a linear system of multiple right-hand sides op(A)*X=B.

Declaration

public static cusolverStatus cusolverDnZgetrs(cusolverDnHandle handle, Operation trans, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr devIpiv, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of right-hand sides.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	devIpiv	array of size at least n, containing pivot indices.
CUdeviceptr	B	array of dimension ldb * nrhs with ldb is not less than max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZheevd(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnZheevd(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZheevd_bufferSize(cusolverDnHandle, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix A.

Declaration

public static cusolverStatus cusolverDnZheevd_bufferSize(cusolverDnHandle handle, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work.

Returns

Type	Description
cusolverStatus

cusolverDnZhegvd(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes eigenvalues and eigenvectors of a symmetric (Hermitian) n×n matrix-pair (A,B).

Declaration

public static cusolverStatus cusolverDnZhegvd(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of work, returned by sygvd_bufferSize.
CUdeviceptr	info	if devInfo = 0, the operation is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i (>0), devInfo indicates either potrf or syevd is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZhegvd_bufferSize(cusolverDnHandle, cusolverEigType, cusolverEigMode, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

The helper functions below can calculate the sizes needed for pre-allocated buffer.

Declaration

public static cusolverStatus cusolverDnZhegvd_bufferSize(cusolverDnHandle handle, cusolverEigType itype, cusolverEigMode jobz, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr W, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
cusolverEigType	itype	Specifies the problem type to be solved.
cusolverEigMode	jobz	specifies options to either compute eigenvalue only or compute eigen-pair.
FillMode	uplo	specifies which part of A and B are stored.
System.Int32	n	number of rows (or columns) of matrix A and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda is not less than max(1,n).
CUdeviceptr	B	array of dimension ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B. ldb is not less than max(1,n).
CUdeviceptr	W	a real array of dimension n. The eigenvalue values of A, sorted so that W(i) >= W(i+1).
System.Int32	lwork	size of work

Returns

Type	Description
cusolverStatus

cusolverDnZhetrd(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

Declaration

public static cusolverStatus cusolverDnZhetrd(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr d, CUdeviceptr e, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle
FillMode	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	d
CUdeviceptr	e
CUdeviceptr	tau
CUdeviceptr	work
System.Int32	lwork
CUdeviceptr	info

Returns

Type	Description
cusolverStatus

cusolverDnZhetrd_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, CUdeviceptr, ref Int32)

Declaration

public static cusolverStatus cusolverDnZhetrd_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr d, CUdeviceptr e, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle
FillMode	uplo
System.Int32	n
CUdeviceptr	A
System.Int32	lda
CUdeviceptr	d
CUdeviceptr	e
CUdeviceptr	tau
System.Int32	lwork

Returns

Type	Description
cusolverStatus

cusolverDnZlaswp(cusolverDnHandle, Int32, CUdeviceptr, Int32, Int32, Int32, CUdeviceptr, Int32)

Declaration

public static cusolverStatus cusolverDnZlaswp(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, int k1, int k2, CUdeviceptr devIpiv, int incx)

Parameters

Type	Name	Description
cusolverDnHandle	handle
System.Int32	n
CUdeviceptr	A
System.Int32	lda
System.Int32	k1
System.Int32	k2
CUdeviceptr	devIpiv
System.Int32	incx

Returns

Type	Description
cusolverStatus

cusolverDnZpotrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnZpotrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr Workspace, int Lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	Workspace	working space, array of size Lwork.
System.Int32	Lwork	size of Workspace
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the leading minor of order i is not positive definite.

Returns

Type	Description
cusolverStatus

cusolverDnZpotrf_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Cholesky factorization of a Hermitian positive-definite matrix.

Declaration

public static cusolverStatus cusolverDnZpotrf_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of Workspace

Returns

Type	Description
cusolverStatus

cusolverDnZpotrs(cusolverDnHandle, FillMode, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function solves a system of linear equations A*X=B where A is a n×n Hermitian matrix, only lower or upper part is meaningful.

Declaration

public static cusolverStatus cusolverDnZpotrs(cusolverDnHandle handle, FillMode uplo, int n, int nrhs, CUdeviceptr A, int lda, CUdeviceptr B, int ldb, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
System.Int32	nrhs	number of columns of matrix X and B.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n). A is either lower cholesky factor L or upper Cholesky factor U.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	B	array of dimension ldb * nrhs. ldb is not less than max(1,n). As an input, B is right hand side matrix. As an output, B is the solution matrix.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CUdeviceptr	devInfo	if devInfo = 0, the Cholesky factorization is successful. if devInfo = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZsytrf(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnZsytrf(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr ipiv, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	ipiv	array of size at least n, containing pivot indices.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working space work.
CUdeviceptr	devInfo	if devInfo = 0, the LU factorization is successful. if devInfo = -i, the i-th parameter is wrong. if devInfo = i, the D(i,i) = 0.

Returns

Type	Description
cusolverStatus

cusolverDnZsytrf_bufferSize(cusolverDnHandle, Int32, CUdeviceptr, Int32, ref Int32)

This function computes the Bunch-Kaufman factorization of a n×n symmetric indefinite matrix.

Declaration

public static cusolverStatus cusolverDnZsytrf_bufferSize(cusolverDnHandle handle, int n, CUdeviceptr A, int lda, ref int Lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	n	number of rows and columns of matrix A.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Int32	Lwork	size of working space work.

Returns

Type	Description
cusolverStatus

cusolverDnZungbr(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnZungbr(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZungbr_bufferSize(cusolverDnHandle, SideMode, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generates one of the unitary matrices Q or P**T determined by GEBRD

Declaration

public static cusolverStatus cusolverDnZungbr_bufferSize(cusolverDnHandle handle, SideMode side, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	if side = CUBLAS_SIDE_LEFT, generate Q. if side = CUBLAS_SIDE_RIGHT, generate P**T.
System.Int32	m	number of rows of matrix Q or P**T.
System.Int32	n	if side = CUBLAS_SIDE_LEFT, m>= n>= min(m,k). if side = CUBLAS_SIDE_RIGHT, n>= m>= min(n,k).
System.Int32	k	if side = CUBLAS_SIDE_LEFT, the number of columns in the original mby- k matrix reduced by gebrd. if side = CUBLAS_SIDE_RIGHT, the number of rows in the original k-by-n matrix reduced by gebrd.
CUdeviceptr	A	array of dimension lda * n On entry, the vectors which define the elementary reflectors, as returned by gebrd. On exit, the m-by-n matrix Q or P**T.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m);
CUdeviceptr	tau	array of dimension min(m,k) if side is CUBLAS_SIDE_LEFT; of dimension min(n,k) if side is CUBLAS_SIDE_RIGHT; tau(i) must contain the scalar factor of the elementary reflector H(i) or G(i), which determines Q or P**T, as returned by gebrd in its array argument TAUQ or TAUP.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnZungqr(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnZungqr(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, rray of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZungqr_bufferSize(cusolverDnHandle, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary matrix Q from QR factorization

Declaration

public static cusolverStatus cusolverDnZungqr_bufferSize(cusolverDnHandle handle, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
System.Int32	m	number of rows of matrix Q. m >= 0;
System.Int32	n	number of columns of matrix Q. m >= n >= 0;
System.Int32	k	number of elementary relfections whose product defines the matrix Q. n >= k >= 0;
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m). i-th column of A contains elementary reflection vector.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,m).
CUdeviceptr	tau	array of dimension k. tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnZungtr(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnZungtr(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the orgtr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZungtr_bufferSize(cusolverDnHandle, FillMode, Int32, CUdeviceptr, Int32, CUdeviceptr, ref Int32)

generate unitary Q comes from sytrd

Declaration

public static cusolverStatus cusolverDnZungtr_bufferSize(cusolverDnHandle handle, FillMode uplo, int n, CUdeviceptr A, int lda, CUdeviceptr tau, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
FillMode	uplo	uplo = CUBLAS_FILL_MODE_LOWER: Lower triangle of A contains elementary reflectors from sytrd. uplo = CUBLAS_FILL_MODE_UPPER: Upper triangle of A contains elementary reflectors from sytrd.
System.Int32	n	number of rows (columns) of matrix Q.
CUdeviceptr	A	array of dimension lda * n On entry, matrix A from sytrd contains the elementary reflectors. On exit, matrix A contains the n-by-n orthogonal matrix Q.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. lda >= max(1,n).
CUdeviceptr	tau	array of dimension (n-1) tau(i) is the scalar of i-th elementary reflection vector.
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus

cusolverDnZunmqr(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnZunmqr(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr devInfo)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * n with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n). The vector tau is from geqrf, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	devInfo	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZunmqr_bufferSize(cusolverDnHandle, SideMode, Operation, Int32, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function computes the QR factorization of a m×n matrix A=Q*R where A is a m×n matrix, Q is a m×n matrix, and R is a n×n upper triangular matrix.

Declaration

public static cusolverStatus cusolverDnZunmqr_bufferSize(cusolverDnHandle handle, SideMode side, Operation trans, int m, int n, int k, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	indicates if matrix Q is on the left or right of C.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
System.Int32	k	number of elementary relfections.
CUdeviceptr	A	array of dimension lda * k with lda is not less than max(1,m).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	tau	array of dimension at least min(m,n).
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)*C.
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array Workspace.

Returns

Type	Description
cusolverStatus

cusolverDnZunmtr(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, CUdeviceptr, Int32, CUdeviceptr)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnZunmtr(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, CUdeviceptr work, int lwork, CUdeviceptr info)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
CUdeviceptr	work	working space, array of size lwork.
System.Int32	lwork	size of working array work.
CUdeviceptr	info	if info = 0, the ormqr is successful. if info = -i, the i-th parameter is wrong.

Returns

Type	Description
cusolverStatus

cusolverDnZunmtr_bufferSize(cusolverDnHandle, SideMode, FillMode, Operation, Int32, Int32, CUdeviceptr, Int32, CUdeviceptr, CUdeviceptr, Int32, ref Int32)

This function overwrites m×n matrix C by

Declaration

public static cusolverStatus cusolverDnZunmtr_bufferSize(cusolverDnHandle handle, SideMode side, FillMode uplo, Operation trans, int m, int n, CUdeviceptr A, int lda, CUdeviceptr tau, CUdeviceptr C, int ldc, ref int lwork)

Parameters

Type	Name	Description
cusolverDnHandle	handle	handle to the cuSolverDN library context.
SideMode	side	side = CUBLAS_SIDE_LEFT, apply Q or QT from the Left; side = CUBLAS_SIDE_RIGHT, apply Q or QT from the Right.
FillMode	uplo	specifies which part of A and B are stored.
Operation	trans	operation op(Q) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix C.
System.Int32	n	number of columns of matrix C.
CUdeviceptr	A	array of dimension lda * m if side = CUBLAS_SIDE_LEFT; lda * n if side = CUBLAS_SIDE_RIGHT. The matrix A from sytrd contains the elementary reflectors.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A. if side is CUBLAS_SIDE_LEFT, lda >= max(1,m); if side is CUBLAS_SIDE_RIGHT, lda >= max(1,n).
CUdeviceptr	tau	array of dimension (m-1) if side is CUBLAS_SIDE_LEFT; of dimension (n-1) if side is CUBLAS_SIDE_RIGHT; The vector tau is from sytrd, so tau(i) is the scalar of i-th elementary reflection vector.
CUdeviceptr	C	array of size ldc * n. On exit, C is overwritten by op(Q)C or Cop(Q).
System.Int32	ldc	leading dimension of two-dimensional array of matrix C. ldc >= max(1,m).
System.Int32	lwork	size of working array work.

Returns

Type	Description
cusolverStatus