Class CudaBlas

Wrapper for CUBLAS

Inheritance

System.Object

CudaBlas

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.CudaBlas

Assembly: CudaBlas.dll

Syntax

public class CudaBlas

Constructors

CudaBlas()

Creates a new cudaBlas handler

Declaration

public CudaBlas()

CudaBlas(CUstream)

Creates a new cudaBlas handler

Declaration

public CudaBlas(CUstream stream)

Parameters

Type	Name	Description
CUstream	stream

CudaBlas(CUstream, AtomicsMode)

Creates a new cudaBlas handler

Declaration

public CudaBlas(CUstream stream, AtomicsMode atomicsmode)

Parameters

Type	Name	Description
CUstream	stream
AtomicsMode	atomicsmode

CudaBlas(CUstream, PointerMode)

Creates a new cudaBlas handler

Declaration

public CudaBlas(CUstream stream, PointerMode pointermode)

Parameters

Type	Name	Description
CUstream	stream
PointerMode	pointermode

CudaBlas(CUstream, PointerMode, AtomicsMode)

Creates a new cudaBlas handler

Declaration

public CudaBlas(CUstream stream, PointerMode pointermode, AtomicsMode atomicsmode)

Parameters

Type	Name	Description
CUstream	stream
PointerMode	pointermode
AtomicsMode	atomicsmode

CudaBlas(AtomicsMode)

Creates a new cudaBlas handler

Declaration

public CudaBlas(AtomicsMode atomicsmode)

Parameters

Type	Name	Description
AtomicsMode	atomicsmode

CudaBlas(PointerMode)

Creates a new cudaBlas handler

Declaration

public CudaBlas(PointerMode pointermode)

Parameters

Type	Name	Description
PointerMode	pointermode

CudaBlas(PointerMode, AtomicsMode)

Creates a new cudaBlas handler

Declaration

public CudaBlas(PointerMode pointermode, AtomicsMode atomicsmode)

Parameters

Type	Name	Description
PointerMode	pointermode
AtomicsMode	atomicsmode

Properties

AtomicsMode

Declaration

public AtomicsMode AtomicsMode { get; set; }

Property Value

Type	Description
AtomicsMode

CublasHandle

Returns the wrapped cublas handle

Declaration

public CudaBlasHandle CublasHandle { get; }

Property Value

Type	Description
CudaBlasHandle

MathMode

Declaration

public Math MathMode { get; set; }

Property Value

Type	Description
Math

PointerMode

Declaration

public PointerMode PointerMode { get; set; }

Property Value

Type	Description
PointerMode

Stream

Declaration

public CUstream Stream { get; set; }

Property Value

Type	Description
CUstream

Methods

AbsoluteSum(CudaDeviceVariable<cuDoubleComplex>, Int32)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public double AbsoluteSum(CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Returns

Type	Description
System.Double

AbsoluteSum(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Double>)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<double> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	result

AbsoluteSum(CudaDeviceVariable<cuDoubleComplex>, Int32, ref Double)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<cuDoubleComplex> x, int incx, ref double result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
System.Double	result

AbsoluteSum(CudaDeviceVariable<cuFloatComplex>, Int32)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public float AbsoluteSum(CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Returns

Type	Description
System.Single

AbsoluteSum(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Single>)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<float> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	result

AbsoluteSum(CudaDeviceVariable<cuFloatComplex>, Int32, ref Single)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<cuFloatComplex> x, int incx, ref float result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
System.Single	result

AbsoluteSum(CudaDeviceVariable<Double>, Int32)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public double AbsoluteSum(CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx

Returns

Type	Description
System.Double

AbsoluteSum(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	result

AbsoluteSum(CudaDeviceVariable<Double>, Int32, ref Double)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<double> x, int incx, ref double result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
System.Double	result

AbsoluteSum(CudaDeviceVariable<Single>, Int32)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public float AbsoluteSum(CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx

Returns

Type	Description
System.Single

AbsoluteSum(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	result

AbsoluteSum(CudaDeviceVariable<Single>, Int32, ref Single)

This function computes the sum of the absolute values of the elements of vector x.

Declaration

public void AbsoluteSum(CudaDeviceVariable<float> x, int incx, ref float result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
System.Single	result

Axpy(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	alpha
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy

Axpy(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	alpha
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy

Axpy(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	alpha
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy

Axpy(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	alpha
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy

Axpy(cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
cuDoubleComplex	alpha
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy

Axpy(cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
cuFloatComplex	alpha
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy

Axpy(Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(double alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
System.Double	alpha
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy

Axpy(Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function multiplies the vector x by the scalar and adds it to the vector y overwriting the latest vector with the result.

Declaration

public void Axpy(float alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
System.Single	alpha
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy

Copy(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy

Copy(CudaDeviceVariable<cuDoubleReal>, Int32, CudaDeviceVariable<cuDoubleReal>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<cuDoubleReal> x, int incx, CudaDeviceVariable<cuDoubleReal> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleReal>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleReal>	y
System.Int32	incy

Copy(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy

Copy(CudaDeviceVariable<cuFloatReal>, Int32, CudaDeviceVariable<cuFloatReal>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<cuFloatReal> x, int incx, CudaDeviceVariable<cuFloatReal> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatReal>	x
System.Int32	incx
CudaDeviceVariable<cuFloatReal>	y
System.Int32	incy

Copy(CudaDeviceVariable<double1>, Int32, CudaDeviceVariable<double1>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<double1> x, int incx, CudaDeviceVariable<double1> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<double1>	x
System.Int32	incx
CudaDeviceVariable<double1>	y
System.Int32	incy

Copy(CudaDeviceVariable<float1>, Int32, CudaDeviceVariable<float1>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<float1> x, int incx, CudaDeviceVariable<float1> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<float1>	x
System.Int32	incx
CudaDeviceVariable<float1>	y
System.Int32	incy

Copy(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy

Copy(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function copies the vector x into the vector y.

Declaration

public void Copy(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy

Ctpttr(FillMode, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the conversion from the triangular packed format to the triangular format.

If uplo == CUBLAS_FILL_MODE_LOWER then the elements of AP are copied into the lower triangular part of the triangular matrix A and the upper part of A is left untouched.

If uplo == CUBLAS_FILL_MODE_UPPER then the elements of AP are copied into the upper triangular part of the triangular matrix A and the lower part of A is left untouched.

Declaration

public void Ctpttr(FillMode uplo, int n, CudaDeviceVariable<cuFloatComplex> AP, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix AP contains lower or upper part of matrix A.
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	AP	array with A stored in packed format.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda x n , with lda>=max(1,n). The opposite side of A is left untouched.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Ctrttp(FillMode, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>)

This function performs the conversion from the triangular format to the triangular packed format.

If uplo == CUBLAS_FILL_MODE_LOWER then the lower triangular part of the triangular matrix A is copied into the array AP.

If uplo == CUBLAS_FILL_MODE_UPPER then then the upper triangular part of the triangular matrix A is copied into the array AP

Declaration

public void Ctrttp(FillMode uplo, int n, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates which matrix A lower or upper part is referenced
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda x n , with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	AP	array with A stored in packed format.

Dgmm(SideMode, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-matrix multiplication C = A x diag(X) if mode == CUBLAS_SIDE_RIGHT, or C = diag(X) x A if mode == CUBLAS_SIDE_LEFT.

where A and C are matrices stored in column-major format with dimensions m*n. X is a vector of size n if mode == CUBLAS_SIDE_RIGHT and of size m if mode == CUBLAS_SIDE_LEFT. X is gathered from one-dimensional array x with stride incx. The absolute value of incx is the stride and the sign of incx is direction of the stride. If incx is positive, then we forward x from the first element. Otherwise, we backward x from the last element.

Declaration

public void Dgmm(SideMode mode, int m, int n, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> X, int incx, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	mode	left multiply if mode == CUBLAS_SIDE_LEFT or right multiply if mode == CUBLAS_SIDE_RIGHT
System.Int32	m	number of rows of matrix A and C.
System.Int32	n	number of columns of matrix A and C.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda x n with lda >= max(1,m)
System.Int32	lda	leading dimension of two-dimensional array used to store the matrix A.
CudaDeviceVariable<cuDoubleComplex>	X	one-dimensional array of size |incx|*m if mode == CUBLAS_SIDE_LEFT and |incx|*n if mode == CUBLAS_SIDE_RIGHT
System.Int32	incx	stride of one-dimensional array x.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc*n with ldc >= max(1,m).
System.Int32	ldc	leading dimension of a two-dimensional array used to store the matrix C.

Dgmm(SideMode, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-matrix multiplication C = A x diag(X) if mode == CUBLAS_SIDE_RIGHT, or C = diag(X) x A if mode == CUBLAS_SIDE_LEFT.

where A and C are matrices stored in column-major format with dimensions m*n. X is a vector of size n if mode == CUBLAS_SIDE_RIGHT and of size m if mode == CUBLAS_SIDE_LEFT. X is gathered from one-dimensional array x with stride incx. The absolute value of incx is the stride and the sign of incx is direction of the stride. If incx is positive, then we forward x from the first element. Otherwise, we backward x from the last element.

Declaration

public void Dgmm(SideMode mode, int m, int n, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> X, int incx, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	mode	left multiply if mode == CUBLAS_SIDE_LEFT or right multiply if mode == CUBLAS_SIDE_RIGHT
System.Int32	m	number of rows of matrix A and C.
System.Int32	n	number of columns of matrix A and C.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda x n with lda >= max(1,m)
System.Int32	lda	leading dimension of two-dimensional array used to store the matrix A.
CudaDeviceVariable<cuFloatComplex>	X	one-dimensional array of size |incx|*m if mode == CUBLAS_SIDE_LEFT and |incx|*n if mode == CUBLAS_SIDE_RIGHT
System.Int32	incx	stride of one-dimensional array x.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc*n with ldc >= max(1,m).
System.Int32	ldc	leading dimension of a two-dimensional array used to store the matrix C.

Dgmm(SideMode, Int32, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-matrix multiplication C = A x diag(X) if mode == CUBLAS_SIDE_RIGHT, or C = diag(X) x A if mode == CUBLAS_SIDE_LEFT.

where A and C are matrices stored in column-major format with dimensions m*n. X is a vector of size n if mode == CUBLAS_SIDE_RIGHT and of size m if mode == CUBLAS_SIDE_LEFT. X is gathered from one-dimensional array x with stride incx. The absolute value of incx is the stride and the sign of incx is direction of the stride. If incx is positive, then we forward x from the first element. Otherwise, we backward x from the last element.

Declaration

public void Dgmm(SideMode mode, int m, int n, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> X, int incx, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
SideMode	mode	left multiply if mode == CUBLAS_SIDE_LEFT or right multiply if mode == CUBLAS_SIDE_RIGHT
System.Int32	m	number of rows of matrix A and C.
System.Int32	n	number of columns of matrix A and C.
CudaDeviceVariable<System.Double>	A	array of dimensions lda x n with lda >= max(1,m)
System.Int32	lda	leading dimension of two-dimensional array used to store the matrix A.
CudaDeviceVariable<System.Double>	X	one-dimensional array of size |incx|*m if mode == CUBLAS_SIDE_LEFT and |incx|*n if mode == CUBLAS_SIDE_RIGHT
System.Int32	incx	stride of one-dimensional array x.
CudaDeviceVariable<System.Double>	C	array of dimensions ldc*n with ldc >= max(1,m).
System.Int32	ldc	leading dimension of a two-dimensional array used to store the matrix C.

Dgmm(SideMode, Int32, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-matrix multiplication C = A x diag(X) if mode == CUBLAS_SIDE_RIGHT, or C = diag(X) x A if mode == CUBLAS_SIDE_LEFT.

where A and C are matrices stored in column-major format with dimensions m*n. X is a vector of size n if mode == CUBLAS_SIDE_RIGHT and of size m if mode == CUBLAS_SIDE_LEFT. X is gathered from one-dimensional array x with stride incx. The absolute value of incx is the stride and the sign of incx is direction of the stride. If incx is positive, then we forward x from the first element. Otherwise, we backward x from the last element.

Declaration

public void Dgmm(SideMode mode, int m, int n, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> X, int incx, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
SideMode	mode	left multiply if mode == CUBLAS_SIDE_LEFT or right multiply if mode == CUBLAS_SIDE_RIGHT
System.Int32	m	number of rows of matrix A and C.
System.Int32	n	number of columns of matrix A and C.
CudaDeviceVariable<System.Single>	A	array of dimensions lda x n with lda >= max(1,m)
System.Int32	lda	leading dimension of two-dimensional array used to store the matrix A.
CudaDeviceVariable<System.Single>	X	one-dimensional array of size |incx|*m if mode == CUBLAS_SIDE_LEFT and |incx|*n if mode == CUBLAS_SIDE_RIGHT
System.Int32	incx	stride of one-dimensional array x.
CudaDeviceVariable<System.Single>	C	array of dimensions ldc*n with ldc >= max(1,m).
System.Int32	ldc	leading dimension of a two-dimensional array used to store the matrix C.

Dispose()

Dispose

Declaration

public void Dispose()

Dispose(Boolean)

For IDisposable

Declaration

protected virtual void Dispose(bool fDisposing)

Parameters

Type	Name	Description
System.Boolean	fDisposing

Dot(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function computes the dot product of vectors x and y.

Declaration

public cuDoubleComplex Dot(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy

Returns

Type	Description
cuDoubleComplex

Dot(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
CudaDeviceVariable<cuDoubleComplex>	result

Dot(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, ref cuDoubleComplex)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, ref cuDoubleComplex result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
cuDoubleComplex	result

Dot(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function computes the dot product of vectors x and y.

Declaration

public cuFloatComplex Dot(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy

Returns

Type	Description
cuFloatComplex

Dot(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
CudaDeviceVariable<cuFloatComplex>	result

Dot(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, ref cuFloatComplex)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, ref cuFloatComplex result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
cuFloatComplex	result

Dot(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function computes the dot product of vectors x and y.

Declaration

public double Dot(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy

Returns

Type	Description
System.Double

Dot(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy
CudaDeviceVariable<System.Double>	result

Dot(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, ref Double)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, ref double result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy
System.Double	result

Dot(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function computes the dot product of vectors x and y.

Declaration

public float Dot(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy

Returns

Type	Description
System.Single

Dot(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy
CudaDeviceVariable<System.Single>	result

Dot(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, ref Single)

This function computes the dot product of vectors x and y.

Declaration

public void Dot(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, ref float result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy
System.Single	result

DotConj(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function computes the dot product of vectors x and y.

Notice that the conjugate of the element of vector x should be used.

Declaration

public cuDoubleComplex DotConj(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy

Returns

Type	Description
cuDoubleComplex

DotConj(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>)

This function computes the dot product of vectors x and y.

Notice that the conjugate of the element of vector x should be used.

Declaration

public void DotConj(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
CudaDeviceVariable<cuDoubleComplex>	result

DotConj(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, ref cuDoubleComplex)

This function computes the dot product of vectors x and y.

Notice that the conjugate of the element of vector x should be used.

Declaration

public void DotConj(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, ref cuDoubleComplex result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
cuDoubleComplex	result

DotConj(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function computes the dot product of vectors x and y.

Notice that the conjugate of the element of vector x should be used.

Declaration

public cuFloatComplex DotConj(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy

Returns

Type	Description
cuFloatComplex

DotConj(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>)

This function computes the dot product of vectors x and y.

Notice that the conjugate of the element of vector x should be used.

Declaration

public void DotConj(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
CudaDeviceVariable<cuFloatComplex>	result

DotConj(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, ref cuFloatComplex)

This function computes the dot product of vectors x and y.

Notice that the conjugate of the element of vector x should be used.

Declaration

public void DotConj(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, ref cuFloatComplex result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
cuFloatComplex	result

Dtpttr(FillMode, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the conversion from the triangular packed format to the triangular format.

If uplo == CUBLAS_FILL_MODE_LOWER then the elements of AP are copied into the lower triangular part of the triangular matrix A and the upper part of A is left untouched.

If uplo == CUBLAS_FILL_MODE_UPPER then the elements of AP are copied into the upper triangular part of the triangular matrix A and the lower part of A is left untouched.

Declaration

public void Dtpttr(FillMode uplo, int n, CudaDeviceVariable<double> AP, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix AP contains lower or upper part of matrix A.
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<System.Double>	AP	array with A stored in packed format.
CudaDeviceVariable<System.Double>	A	array of dimensions lda x n , with lda>=max(1,n). The opposite side of A is left untouched.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Dtrttp(FillMode, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Single>)

This function performs the conversion from the triangular format to the triangular packed format.

If uplo == CUBLAS_FILL_MODE_LOWER then the lower triangular part of the triangular matrix A is copied into the array AP.

If uplo == CUBLAS_FILL_MODE_UPPER then then the upper triangular part of the triangular matrix A is copied into the array AP

Declaration

public void Dtrttp(FillMode uplo, int n, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<float> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates which matrix A lower or upper part is referenced
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<System.Double>	A	array of dimensions lda x n , with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	AP	array with A stored in packed format.

Finalize()

For dispose

Declaration

protected void Finalize()

Gbmv(Operation, Int32, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gbmv(Operation, Int32, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gbmv(Operation, Int32, Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gbmv(Operation, Int32, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gbmv(Operation, Int32, Int32, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuDoubleComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gbmv(Operation, Int32, Int32, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuFloatComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gbmv(Operation, Int32, Int32, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, double beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Double	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gbmv(Operation, Int32, Int32, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gbmv(Operation trans, int m, int n, int kl, int ku, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, float beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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	kl	number of subdiagonals of matrix A.
System.Int32	ku	number of superdiagonals of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Single	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Geam(Operation, Operation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Geam(Operation, Operation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Geam(Operation, Operation, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Geam(Operation, Operation, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Geam(Operation, Operation, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuDoubleComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Geam(Operation, Operation, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuFloatComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Geam(Operation, Operation, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Double	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Geam(Operation, Operation, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-matrix addition/transposition C = alpha * Op(A) + beta * Op(B) where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mn, op(B) mn and C m*n, respectively.

Declaration

public void Geam(Operation transa, Operation transb, int m, int n, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

GelsBatchedC(Operation, Int32, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>)

This function find the least squares solution of a batch of overdetermined systems. On exit, each Aarray[i] is overwritten with their QR factorization and each Carray[i] is overwritten with the least square solution GelsBatched supports only the non-transpose operation and only solves overdetermined systems (m >= n).

GelsBatched only supports compute capability 2.0 or above.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GelsBatchedC(Operation trans, int m, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> devInfoArray)

Parameters

Type	Name	Description
Operation	trans	operation op(Aarray[i]) that is non- or (conj.) transpose. Only non-transpose operation is currently supported.
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of each Aarray[i] and rows of each Carray[i].
System.Int32	nrhs	number of columns of each Carray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i]
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to device array, with each array of dim. m x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	devInfoArray	null or optional array of integers of dimension batchsize.

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

GelsBatchedD(Operation, Int32, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>)

This function find the least squares solution of a batch of overdetermined systems. On exit, each Aarray[i] is overwritten with their QR factorization and each Carray[i] is overwritten with the least square solution GelsBatched supports only the non-transpose operation and only solves overdetermined systems (m >= n).

GelsBatched only supports compute capability 2.0 or above.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GelsBatchedD(Operation trans, int m, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> devInfoArray)

Parameters

Type	Name	Description
Operation	trans	operation op(Aarray[i]) that is non- or (conj.) transpose. Only non-transpose operation is currently supported.
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of each Aarray[i] and rows of each Carray[i].
System.Int32	nrhs	number of columns of each Carray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i]
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to device array, with each array of dim. m x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	devInfoArray	null or optional array of integers of dimension batchsize.

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

GelsBatchedS(Operation, Int32, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>)

This function find the least squares solution of a batch of overdetermined systems. On exit, each Aarray[i] is overwritten with their QR factorization and each Carray[i] is overwritten with the least square solution GelsBatched supports only the non-transpose operation and only solves overdetermined systems (m >= n).

GelsBatched only supports compute capability 2.0 or above.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GelsBatchedS(Operation trans, int m, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> devInfoArray)

Parameters

Type	Name	Description
Operation	trans	operation op(Aarray[i]) that is non- or (conj.) transpose. Only non-transpose operation is currently supported.
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of each Aarray[i] and rows of each Carray[i].
System.Int32	nrhs	number of columns of each Carray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i]
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to device array, with each array of dim. m x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	devInfoArray	null or optional array of integers of dimension batchsize.

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

GelsBatchedZ(Operation, Int32, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>)

This function find the least squares solution of a batch of overdetermined systems. On exit, each Aarray[i] is overwritten with their QR factorization and each Carray[i] is overwritten with the least square solution GelsBatched supports only the non-transpose operation and only solves overdetermined systems (m >= n).

GelsBatched only supports compute capability 2.0 or above.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GelsBatchedZ(Operation trans, int m, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> devInfoArray)

Parameters

Type	Name	Description
Operation	trans	operation op(Aarray[i]) that is non- or (conj.) transpose. Only non-transpose operation is currently supported.
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of each Aarray[i] and rows of each Carray[i].
System.Int32	nrhs	number of columns of each Carray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i]
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to device array, with each array of dim. m x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	devInfoArray	null or optional array of integers of dimension batchsize.

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

Gemm(Operation, Operation, Int32, Int32, Int32, half, CudaDeviceVariable<half>, Int32, CudaDeviceVariable<half>, Int32, half, CudaDeviceVariable<half>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, half alpha, CudaDeviceVariable<half> A, int lda, CudaDeviceVariable<half> B, int ldb, half beta, CudaDeviceVariable<half> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
half	alpha	scalar used for multiplication.
CudaDeviceVariable<half>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<half>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
half	beta	scalar used for multiplication.
CudaDeviceVariable<half>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<half>, CudaDeviceVariable<half>, Int32, CudaDeviceVariable<half>, Int32, CudaDeviceVariable<half>, CudaDeviceVariable<half>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<half> alpha, CudaDeviceVariable<half> A, int lda, CudaDeviceVariable<half> B, int ldb, CudaDeviceVariable<half> beta, CudaDeviceVariable<half> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<half>	alpha	scalar used for multiplication.
CudaDeviceVariable<half>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<half>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<half>	beta	scalar used for multiplication.
CudaDeviceVariable<half>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuDoubleComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuFloatComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Double	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemm(Operation, Operation, Int32, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void Gemm(Operation transa, Operation transb, int m, int n, int k, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, half, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, half, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, half alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, half beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
half	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
half	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<half>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<half>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<half> alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, CudaDeviceVariable<half> beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
CudaDeviceVariable<half>	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
CudaDeviceVariable<half>	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<double> alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<CUdeviceptr>, cudaDataType, Int32, CudaDeviceVariable<CUdeviceptr>, cudaDataType, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<CUdeviceptr>, cudaDataType, Int32, Int32, cudaDataType, GemmAlgo)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<CUdeviceptr> Aarray, cudaDataType Atype, int lda, CudaDeviceVariable<CUdeviceptr> Barray, cudaDataType Btype, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<CUdeviceptr> Carray, cudaDataType Ctype, int ldc, int batchCount, cudaDataType computeType, GemmAlgo algo)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
cudaDataType	Atype
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
cudaDataType	Btype
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
cudaDataType	Ctype
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.
cudaDataType	computeType
GemmAlgo	algo

GemmBatched(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, cuDoubleComplex, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, cuDoubleComplex alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, cuDoubleComplex beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
cuDoubleComplex	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, cuFloatComplex, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, cuFloatComplex, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, cuFloatComplex alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, cuFloatComplex beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
cuFloatComplex	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, Double, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Double, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, double alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, double beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
System.Double	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmBatched(Operation, Operation, Int32, Int32, Int32, IntPtr, CudaDeviceVariable<CUdeviceptr>, cudaDataType, Int32, CudaDeviceVariable<CUdeviceptr>, cudaDataType, Int32, IntPtr, CudaDeviceVariable<CUdeviceptr>, cudaDataType, Int32, Int32, cudaDataType, GemmAlgo)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, IntPtr alpha, CudaDeviceVariable<CUdeviceptr> Aarray, cudaDataType Atype, int lda, CudaDeviceVariable<CUdeviceptr> Barray, cudaDataType Btype, int ldb, IntPtr beta, CudaDeviceVariable<CUdeviceptr> Carray, cudaDataType Ctype, int ldc, int batchCount, cudaDataType computeType, GemmAlgo algo)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
System.IntPtr	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
cudaDataType	Atype
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
cudaDataType	Btype
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
System.IntPtr	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
cudaDataType	Ctype
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.
cudaDataType	computeType
GemmAlgo	algo

GemmBatched(Operation, Operation, Int32, Int32, Int32, Single, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Single, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function performs the matrix-matrix multiplications of an array of matrices. where and are scalars, and , and are arrays of pointers to matrices stored in column-major format with dimensions op(A[i])m x k, op(B[i])k x n and op(C[i])m x n, respectively.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. For small sizes, typically smaller than 100x100, this function improves significantly performance compared to making calls to its corresponding cublas<type>gemm routine. However, on GPU architectures that support concurrent kernels, it might be advantageous to make multiple calls to cublas<type>gemm into different streams as the matrix sizes increase.

Declaration

public void GemmBatched(Operation transa, Operation transb, int m, int n, int k, float alpha, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, float beta, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, int batchCount)

Parameters

Type	Name	Description
Operation	transa	operation op(A[i]) that is non- or (conj.) transpose.
Operation	transb	operation op(B[i]) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A[i]) and C[i].
System.Int32	n	number of columns of op(B[i]) and C[i].
System.Int32	k	number of columns of op(A[i]) and rows of op(B[i]).
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<CUdeviceptr>	Aarray	array of device pointers, with each array/device pointer of dim. lda x k with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x m with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<CUdeviceptr>	Barray	array of device pointers, with each array of dim. ldb x n with ldb>=max(1,k) if transa==CUBLAS_OP_N and ldb x k with ldb>=max(1,n) max(1,) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store each matrix B[i].
System.Single	beta	scalar used for multiplication. If beta == 0, C does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	Carray	array of device pointers. It has dimensions ldc x n with ldc>=max(1,m).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix C[i].
System.Int32	batchCount	number of pointers contained in A, B and C.

GemmEx(Operation, Operation, Int32, Int32, Int32, CudaDeviceVariable<Single>, CUdeviceptr, DataType, Int32, CUdeviceptr, DataType, Int32, CudaDeviceVariable<Single>, CUdeviceptr, DataType, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void GemmEx(Operation transa, Operation transb, int m, int n, int k, CudaDeviceVariable<float> alpha, CUdeviceptr A, DataType Atype, int lda, CUdeviceptr B, DataType Btype, int ldb, CudaDeviceVariable<float> beta, CUdeviceptr C, DataType Ctype, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CUdeviceptr	A	array of dimensions lda * k.
DataType	Atype	enumerant specifying the datatype of matrix A.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	B	array of dimensions ldb * n.
DataType	Btype	enumerant specifying the datatype of matrix B.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CUdeviceptr	C	array of dimensions ldb * n.
DataType	Ctype	enumerant specifying the datatype of matrix C.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

GemmEx(Operation, Operation, Int32, Int32, Int32, Single, CUdeviceptr, DataType, Int32, CUdeviceptr, DataType, Int32, Single, CUdeviceptr, DataType, Int32)

This function performs the matrix-matrix multiplication C = alpha * Op(A) * Op(B) + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in column-major format with dimensions op(A) mk, op(B) kn and C m*n, respectively.

Declaration

public void GemmEx(Operation transa, Operation transb, int m, int n, int k, float alpha, CUdeviceptr A, DataType Atype, int lda, CUdeviceptr B, DataType Btype, int ldb, float beta, CUdeviceptr C, DataType Ctype, int ldc)

Parameters

Type	Name	Description
Operation	transa	operation op(A) that is non- or (conj.) transpose.
Operation	transb	operation op(B) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix op(A) and C.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Single	alpha	scalar used for multiplication.
CUdeviceptr	A	array of dimensions lda * k.
DataType	Atype	enumerant specifying the datatype of matrix A.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CUdeviceptr	B	array of dimensions ldb * n.
DataType	Btype	enumerant specifying the datatype of matrix B.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	scalar used for multiplication.
CUdeviceptr	C	array of dimensions ldb * n.
DataType	Ctype	enumerant specifying the datatype of matrix C.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Gemv(Operation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gemv(Operation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gemv(Operation, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gemv(Operation, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gemv(Operation, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuDoubleComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gemv(Operation, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	m	number of rows of matrix A.
System.Int32	n	number of columns of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuFloatComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gemv(Operation, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, double beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Double	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Gemv(Operation, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the matrix-vector multiplication y = alpha * Op(A) * x + beta * y where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha and beta are scalars.

Declaration

public void Gemv(Operation trans, int m, int n, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, float beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
Operation	trans	operation op(A) 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.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Single	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

GeqrfBatchedC(Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>)

This function performs the QR factorization of each Aarray[i] for i = 0, ...,batchSize-1 using Householder reflections. Each matrix Q[i] is represented as a product of elementary reflectors and is stored in the lower part of each Aarray[i]. This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

cublas<t>geqrfBatched supports arbitrary dimension.

cublas<t>geqrfBatched only supports compute capability 2.0 or above.

Declaration

public int GeqrfBatchedC(int m, int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> TauArray)

Parameters

Type	Name	Description
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<CUdeviceptr>	TauArray	array of pointers to device vector, with each vector of dim. max(1,min(m,n)).

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

GeqrfBatchedD(Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>)

This function performs the QR factorization of each Aarray[i] for i = 0, ...,batchSize-1 using Householder reflections. Each matrix Q[i] is represented as a product of elementary reflectors and is stored in the lower part of each Aarray[i]. This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

cublas<t>geqrfBatched supports arbitrary dimension.

cublas<t>geqrfBatched only supports compute capability 2.0 or above.

Declaration

public int GeqrfBatchedD(int m, int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> TauArray)

Parameters

Type	Name	Description
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<CUdeviceptr>	TauArray	array of pointers to device vector, with each vector of dim. max(1,min(m,n)).

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

GeqrfBatchedS(Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>)

This function performs the QR factorization of each Aarray[i] for i = 0, ...,batchSize-1 using Householder reflections. Each matrix Q[i] is represented as a product of elementary reflectors and is stored in the lower part of each Aarray[i]. This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

cublas<t>geqrfBatched supports arbitrary dimension.

cublas<t>geqrfBatched only supports compute capability 2.0 or above.

Declaration

public int GeqrfBatchedS(int m, int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> TauArray)

Parameters

Type	Name	Description
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<CUdeviceptr>	TauArray	array of pointers to device vector, with each vector of dim. max(1,min(m,n)).

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

GeqrfBatchedZ(Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>)

This function performs the QR factorization of each Aarray[i] for i = 0, ...,batchSize-1 using Householder reflections. Each matrix Q[i] is represented as a product of elementary reflectors and is stored in the lower part of each Aarray[i]. This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

cublas<t>geqrfBatched supports arbitrary dimension.

cublas<t>geqrfBatched only supports compute capability 2.0 or above.

Declaration

public int GeqrfBatchedZ(int m, int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<CUdeviceptr> TauArray)

Parameters

Type	Name	Description
System.Int32	m	number of rows Aarray[i].
System.Int32	n	number of columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to device array, with each array of dim. m x n with lda>=max(1,m). The array size determines the number of batches.
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<CUdeviceptr>	TauArray	array of pointers to device vector, with each vector of dim. max(1,min(m,n)).

Returns

Type	Description
System.Int32	0, if the parameters passed to the function are valid, <0, if the parameter in postion -value is invalid

Ger(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void Ger(CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Ger(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void Ger(CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Ger(Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void Ger(double alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Ger(Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void Ger(float alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerC(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^H + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerC(CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerC(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^H + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerC(CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerC(cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^H + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerC(cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerC(cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^H + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerC(cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerU(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerU(CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerU(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerU(CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerU(cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerU(cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GerU(cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the rank-1 update A = alpha * x * y^T + A where A is a m*n matrix stored in column-major format, x and y are vectors, and alpha is a scalar. m = x.Size, n = y.Size.

Declaration

public void GerU(cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

GetMatrix<T>(Int32, Int32, CudaDeviceVariable<T>, Int32, T[], Int32)

copies a tile of rows x cols elements from a matrix devSource in GPU memory space to a matrix hostDest in CPU memory space. Both matrices are assumed to be stored in column major format, with the leading dimension (i.e. number of rows) of source matrix devSource provided in devSource, and the leading dimension of matrix hostDest provided in ldHostDest.

Declaration

public static void GetMatrix<T>(int rows, int cols, CudaDeviceVariable<T> devSource, int ldDevSource, T[] hostDest, int ldHostDest)
    where T : struct

Parameters

Type	Name	Description
System.Int32	rows
System.Int32	cols
CudaDeviceVariable<T>	devSource
System.Int32	ldDevSource
T[]	hostDest
System.Int32	ldHostDest

Type Parameters

Name	Description
T

GetMatrixAsync<T>(Int32, Int32, CudaDeviceVariable<T>, Int32, T[], Int32, CUstream)

copies a tile of rows x cols elements from a matrix devSource in GPU memory space to a matrix hostDest in CPU memory space. Both matrices are assumed to be stored in column major format, with the leading dimension (i.e. number of rows) of source matrix devSource provided in devSource, and the leading dimension of matrix hostDest provided in ldHostDest.

Declaration

public static void GetMatrixAsync<T>(int rows, int cols, CudaDeviceVariable<T> devSource, int ldDevSource, T[] hostDest, int ldHostDest, CUstream stream)
    where T : struct

Parameters

Type	Name	Description
System.Int32	rows
System.Int32	cols
CudaDeviceVariable<T>	devSource
System.Int32	ldDevSource
T[]	hostDest
System.Int32	ldHostDest
CUstream	stream

Type Parameters

Name	Description
T

GetrfBatchedC(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32)

This function performs the LU factorization of an array of n x n matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimension n to 32.

Declaration

public void GetrfBatchedC(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of A[i].
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointer of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<System.Int32>	P	array of size n x batchSize that contains the permutation vector of each factorization of A[i] stored in a linear fashion.
CudaDeviceVariable<System.Int32>	INFO	If info=0, the execution is successful. If info = -i, the i-th parameter had an illegal value. If info = i, aii is 0. The factorization has been completed, but U is exactly singular.
System.Int32	batchSize	number of pointers contained in A

GetrfBatchedD(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32)

This function performs the LU factorization of an array of n x n matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimension n to 32.

Declaration

public void GetrfBatchedD(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of A[i].
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointer of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<System.Int32>	P	array of size n x batchSize that contains the permutation vector of each factorization of A[i] stored in a linear fashion.
CudaDeviceVariable<System.Int32>	INFO	If info=0, the execution is successful. If info = -i, the i-th parameter had an illegal value. If info = i, aii is 0. The factorization has been completed, but U is exactly singular.
System.Int32	batchSize	number of pointers contained in A

GetrfBatchedS(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32)

This function performs the LU factorization of an array of n x n matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimension n to 32.

Declaration

public void GetrfBatchedS(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of A[i].
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointer of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<System.Int32>	P	array of size n x batchSize that contains the permutation vector of each factorization of A[i] stored in a linear fashion.
CudaDeviceVariable<System.Int32>	INFO	If info=0, the execution is successful. If info = -i, the i-th parameter had an illegal value. If info = i, aii is 0. The factorization has been completed, but U is exactly singular.
System.Int32	batchSize	number of pointers contained in A

GetrfBatchedZ(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<Int32>, Int32)

This function performs the LU factorization of an array of n x n matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimension n to 32.

Declaration

public void GetrfBatchedZ(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of A[i].
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointer of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix A[i].
CudaDeviceVariable<System.Int32>	P	array of size n x batchSize that contains the permutation vector of each factorization of A[i] stored in a linear fashion.
CudaDeviceVariable<System.Int32>	INFO	If info=0, the execution is successful. If info = -i, the i-th parameter had an illegal value. If info = i, aii is 0. The factorization has been completed, but U is exactly singular.
System.Int32	batchSize	number of pointers contained in A

GetriBatchedC(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Aarray and Carray are arrays of pointers to matrices stored in column-major format with dimensions n*n and leading dimension lda and ldc respectively. This function performs the inversion of matrices A[i] for i = 0, ..., batchSize-1.

Prior to calling GetriBatched, the matrix A[i] must be factorized first using the routine GetrfBatched. After the call of GetrfBatched, the matrix pointing by Aarray[i] will contain the LU factors of the matrix A[i] and the vector pointing by (PivotArray+i) will contain the pivoting sequence.

Following the LU factorization, GetriBatched uses forward and backward triangular solvers to complete inversion of matrices A[i] for i = 0, ..., batchSize-1. The inversion is out-of-place, so memory space of Carray[i] cannot overlap memory space of Array[i].

Declaration

public void GetriBatchedC(int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dimension n*n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	P	array of size n*batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion.
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to array, with each array of dimension n*n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	INFO	array of size batchSize that info(=infoArray[i]) contains the information of inversion of A[i]. If info=0, the execution is successful. If info = k, U(k,k) is 0. The U is exactly singular and the inversion failed.
System.Int32	batchSize	number of pointers contained in A

GetriBatchedD(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Aarray and Carray are arrays of pointers to matrices stored in column-major format with dimensions n*n and leading dimension lda and ldc respectively. This function performs the inversion of matrices A[i] for i = 0, ..., batchSize-1.

Prior to calling GetriBatched, the matrix A[i] must be factorized first using the routine GetrfBatched. After the call of GetrfBatched, the matrix pointing by Aarray[i] will contain the LU factors of the matrix A[i] and the vector pointing by (PivotArray+i) will contain the pivoting sequence.

Following the LU factorization, GetriBatched uses forward and backward triangular solvers to complete inversion of matrices A[i] for i = 0, ..., batchSize-1. The inversion is out-of-place, so memory space of Carray[i] cannot overlap memory space of Array[i].

Declaration

public void GetriBatchedD(int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dimension n*n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	P	array of size n*batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion.
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to array, with each array of dimension n*n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	INFO	array of size batchSize that info(=infoArray[i]) contains the information of inversion of A[i]. If info=0, the execution is successful. If info = k, U(k,k) is 0. The U is exactly singular and the inversion failed.
System.Int32	batchSize	number of pointers contained in A

GetriBatchedS(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Aarray and Carray are arrays of pointers to matrices stored in column-major format with dimensions n*n and leading dimension lda and ldc respectively. This function performs the inversion of matrices A[i] for i = 0, ..., batchSize-1.

Prior to calling GetriBatched, the matrix A[i] must be factorized first using the routine GetrfBatched. After the call of GetrfBatched, the matrix pointing by Aarray[i] will contain the LU factors of the matrix A[i] and the vector pointing by (PivotArray+i) will contain the pivoting sequence.

Following the LU factorization, GetriBatched uses forward and backward triangular solvers to complete inversion of matrices A[i] for i = 0, ..., batchSize-1. The inversion is out-of-place, so memory space of Carray[i] cannot overlap memory space of Array[i].

Declaration

public void GetriBatchedS(int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dimension n*n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	P	array of size n*batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion.
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to array, with each array of dimension n*n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	INFO	array of size batchSize that info(=infoArray[i]) contains the information of inversion of A[i]. If info=0, the execution is successful. If info = k, U(k,k) is 0. The U is exactly singular and the inversion failed.
System.Int32	batchSize	number of pointers contained in A

GetriBatchedZ(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Aarray and Carray are arrays of pointers to matrices stored in column-major format with dimensions n*n and leading dimension lda and ldc respectively. This function performs the inversion of matrices A[i] for i = 0, ..., batchSize-1.

Prior to calling GetriBatched, the matrix A[i] must be factorized first using the routine GetrfBatched. After the call of GetrfBatched, the matrix pointing by Aarray[i] will contain the LU factors of the matrix A[i] and the vector pointing by (PivotArray+i) will contain the pivoting sequence.

Following the LU factorization, GetriBatched uses forward and backward triangular solvers to complete inversion of matrices A[i] for i = 0, ..., batchSize-1. The inversion is out-of-place, so memory space of Carray[i] cannot overlap memory space of Array[i].

Declaration

public void GetriBatchedZ(int n, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> P, CudaDeviceVariable<CUdeviceptr> Carray, int ldc, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n	number of rows and columns of Aarray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dimension n*n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	P	array of size n*batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion.
CudaDeviceVariable<CUdeviceptr>	Carray	array of pointers to array, with each array of dimension n*n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store each matrix Carray[i].
CudaDeviceVariable<System.Int32>	INFO	array of size batchSize that info(=infoArray[i]) contains the information of inversion of A[i]. If info=0, the execution is successful. If info = k, U(k,k) is 0. The U is exactly singular and the inversion failed.
System.Int32	batchSize	number of pointers contained in A

GetrsBatchedC(Operation, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of systems of linear equations of the form:

op(A[i]) X[i] = a B[i]

where A[i] is a matrix which has been LU factorized with pivoting, X[i] and B[i] are n x nrhs matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GetrsBatchedC(Operation trans, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> devIpiv, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, int batchSize)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of Aarray[i].
System.Int32	nrhs	number of columns of Barray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	devIpiv	array of size n x batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion. If devIpiv is nil, pivoting for all Aarray[i] is ignored.
CudaDeviceVariable<CUdeviceptr>	Barray	array of pointers to array, with each array of dim. n x nrhs with ldb>=max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store each solution matrix Barray[i].
System.Int32	batchSize	number of pointers contained in A

Returns

Type	Description
System.Int32	If info=0, the execution is successful. If info = -j, the j-th parameter had an illegal value.

GetrsBatchedD(Operation, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of systems of linear equations of the form:

op(A[i]) X[i] = a B[i]

where A[i] is a matrix which has been LU factorized with pivoting, X[i] and B[i] are n x nrhs matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GetrsBatchedD(Operation trans, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> devIpiv, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, int batchSize)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of Aarray[i].
System.Int32	nrhs	number of columns of Barray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	devIpiv	array of size n x batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion. If devIpiv is nil, pivoting for all Aarray[i] is ignored.
CudaDeviceVariable<CUdeviceptr>	Barray	array of pointers to array, with each array of dim. n x nrhs with ldb>=max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store each solution matrix Barray[i].
System.Int32	batchSize	number of pointers contained in A

Returns

Type	Description
System.Int32	If info=0, the execution is successful. If info = -j, the j-th parameter had an illegal value.

GetrsBatchedS(Operation, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of systems of linear equations of the form:

op(A[i]) X[i] = a B[i]

where A[i] is a matrix which has been LU factorized with pivoting, X[i] and B[i] are n x nrhs matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GetrsBatchedS(Operation trans, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> devIpiv, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, int batchSize)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of Aarray[i].
System.Int32	nrhs	number of columns of Barray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	devIpiv	array of size n x batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion. If devIpiv is nil, pivoting for all Aarray[i] is ignored.
CudaDeviceVariable<CUdeviceptr>	Barray	array of pointers to array, with each array of dim. n x nrhs with ldb>=max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store each solution matrix Barray[i].
System.Int32	batchSize	number of pointers contained in A

Returns

Type	Description
System.Int32	If info=0, the execution is successful. If info = -j, the j-th parameter had an illegal value.

GetrsBatchedZ(Operation, Int32, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of systems of linear equations of the form:

op(A[i]) X[i] = a B[i]

where A[i] is a matrix which has been LU factorized with pivoting, X[i] and B[i] are n x nrhs matrices.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor.

Declaration

public int GetrsBatchedZ(Operation trans, int n, int nrhs, CudaDeviceVariable<CUdeviceptr> Aarray, int lda, CudaDeviceVariable<int> devIpiv, CudaDeviceVariable<CUdeviceptr> Barray, int ldb, int batchSize)

Parameters

Type	Name	Description
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows and columns of Aarray[i].
System.Int32	nrhs	number of columns of Barray[i].
CudaDeviceVariable<CUdeviceptr>	Aarray	array of pointers to array, with each array of dim. n x n with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store each matrix Aarray[i].
CudaDeviceVariable<System.Int32>	devIpiv	array of size n x batchSize that contains the pivoting sequence of each factorization of Aarray[i] stored in a linear fashion. If devIpiv is nil, pivoting for all Aarray[i] is ignored.
CudaDeviceVariable<CUdeviceptr>	Barray	array of pointers to array, with each array of dim. n x nrhs with ldb>=max(1,n).
System.Int32	ldb	leading dimension of two-dimensional array used to store each solution matrix Barray[i].
System.Int32	batchSize	number of pointers contained in A

Returns

Type	Description
System.Int32	If info=0, the execution is successful. If info = -j, the j-th parameter had an illegal value.

GetVector<T>(CudaDeviceVariable<T>, Int32, T[], Int32)

copies elements from a vector devSourceVector in GPU memory space to a vector hostDestVector in CPU memory space. Storage spacing between consecutive elements is incrHostDest for the source vector devSourceVector and incrDevSource for the destination vector hostDestVector. Column major format for two-dimensional matrices is assumed throughout CUBLAS. Therefore, if the increment for a vector is equal to 1, this access a column vector while using an increment equal to the leading dimension of the respective matrix accesses a row vector.

Declaration

public static void GetVector<T>(CudaDeviceVariable<T> devSourceVector, int incrDevSource, T[] hostDestVector, int incrHostDest)
    where T : struct

Parameters

Type	Name	Description
CudaDeviceVariable<T>	devSourceVector	Source vector in device memory
System.Int32	incrDevSource
T[]	hostDestVector	Destination vector in host memory
System.Int32	incrHostDest

Type Parameters

Name	Description
T	Vector datatype

GetVectorAsync<T>(CudaDeviceVariable<T>, Int32, T[], Int32, CUstream)

copies elements from a vector devSourceVector in GPU memory space to a vector hostDestVector in CPU memory space. Storage spacing between consecutive elements is incrHostDest for the source vector devSourceVector and incrDevSource for the destination vector hostDestVector. Column major format for two-dimensional matrices is assumed throughout CUBLAS. Therefore, if the increment for a vector is equal to 1, this access a column vector while using an increment equal to the leading dimension of the respective matrix accesses a row vector.

Declaration

public static void GetVectorAsync<T>(CudaDeviceVariable<T> devSourceVector, int incrDevSource, T[] hostDestVector, int incrHostDest, CUstream stream)
    where T : struct

Parameters

Type	Name	Description
CudaDeviceVariable<T>	devSourceVector	Source vector in device memory
System.Int32	incrDevSource
T[]	hostDestVector	Destination vector in host memory
System.Int32	incrHostDest
CUstream	stream

Type Parameters

Name	Description
T	Vector datatype

GetVersion()

Declaration

public Version GetVersion()

Returns

Type	Description
System.Version

Hbmv(FillMode, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hbmv(FillMode uplo, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hbmv(FillMode, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hbmv(FillMode uplo, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hbmv(FillMode, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hbmv(FillMode uplo, int k, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuDoubleComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hbmv(FillMode, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hbmv(FillMode uplo, int k, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuFloatComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hemm(SideMode, FillMode, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a Hermitian matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Hemm(SideMode side, FillMode uplo, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Hemm(SideMode, FillMode, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a Hermitian matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Hemm(SideMode side, FillMode uplo, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Hemm(SideMode, FillMode, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a Hermitian matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Hemm(SideMode side, FillMode uplo, int m, int n, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuDoubleComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Hemm(SideMode, FillMode, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a Hermitian matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Hemm(SideMode side, FillMode uplo, int m, int n, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuFloatComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Hemv(FillMode, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hemv(FillMode uplo, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hemv(FillMode, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hemv(FillMode uplo, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hemv(FillMode, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hemv(FillMode uplo, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuDoubleComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hemv(FillMode, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hemv(FillMode uplo, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuFloatComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Her(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Her(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Her(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her(FillMode, Double, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Her(FillMode uplo, double alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her(FillMode, Single, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Her(FillMode uplo, float alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her2(FillMode, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Her2(FillMode uplo, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her2(FillMode, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Her2(FillMode uplo, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her2(FillMode, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Her2(FillMode uplo, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her2(FillMode, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Her2(FillMode uplo, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Her2k(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * (Op(A)Op(B)^H + Op(B)Op(A)^H) + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Her2k(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Her2k(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * (Op(A)Op(B)^H + Op(B)Op(A)^H) + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Her2k(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Her2k(FillMode, Operation, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, Double, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * (Op(A)Op(B)^H + Op(B)Op(A)^H) + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Her2k(FillMode uplo, Operation trans, int n, int k, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, double beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Double	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Her2k(FillMode, Operation, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, Single, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * (Op(A)Op(B)^H + Op(B)Op(A)^H) + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Her2k(FillMode uplo, Operation trans, int n, int k, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, float beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herk(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * Op(A)Op(A)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Herk(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<double> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<double> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herk(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * Op(A)Op(A)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Herk(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<float> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herk(FillMode, Operation, Int32, Int32, Double, CudaDeviceVariable<cuDoubleComplex>, Int32, Double, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * Op(A)Op(A)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Herk(FillMode uplo, Operation trans, int n, int k, double alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, double beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Double	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herk(FillMode, Operation, Int32, Int32, Single, CudaDeviceVariable<cuFloatComplex>, Int32, Single, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian rank-k update C = alpha * Op(A)Op(A)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Herk(FillMode uplo, Operation trans, int n, int k, float alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, float beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Single	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herkx(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs a variation of the Hermitian rank-k update C = alpha * Op(A) * Op(B)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Herkx(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other Hermitian part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimension ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	beta	real scalar used for multiplication, if beta==0 then C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimension ldc x n, with ldc>=max(1,n). The imaginary parts of the diagonal elements are assumed and set to zero.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herkx(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs a variation of the Hermitian rank-k update C = alpha * Op(A) * Op(B)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Herkx(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other Hermitian part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimension ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	real scalar used for multiplication, if beta==0 then C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimension ldc x n, with ldc>=max(1,n). The imaginary parts of the diagonal elements are assumed and set to zero.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herkx(FillMode, Operation, Int32, Int32, ref cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, ref Double, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs a variation of the Hermitian rank-k update C = alpha * Op(A) * Op(B)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Herkx(FillMode uplo, Operation trans, int n, int k, ref cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, ref double beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other Hermitian part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimension ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Double	beta	real scalar used for multiplication, if beta==0 then C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimension ldc x n, with ldc>=max(1,n). The imaginary parts of the diagonal elements are assumed and set to zero.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Herkx(FillMode, Operation, Int32, Int32, ref cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, ref Single, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs a variation of the Hermitian rank-k update C = alpha * Op(A) * Op(B)^H + beta * C where alpha and beta are scalars, and C is a Hermitian matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and Op(B) nk, respectively.

Declaration

public void Herkx(FillMode uplo, Operation trans, int n, int k, ref cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, ref float beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other Hermitian part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimension ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	real scalar used for multiplication, if beta==0 then C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimension ldc x n, with ldc>=max(1,n). The imaginary parts of the diagonal elements are assumed and set to zero.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Hpmv(FillMode, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hpmv(FillMode uplo, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> AP, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hpmv(FillMode, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hpmv(FillMode uplo, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> AP, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hpmv(FillMode, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the Hermitian packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hpmv(FillMode uplo, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> AP, CudaDeviceVariable<cuDoubleComplex> x, int incx, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuDoubleComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hpmv(FillMode, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the Hermitian packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n Hermitian matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Hpmv(FillMode uplo, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> AP, CudaDeviceVariable<cuFloatComplex> x, int incx, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuFloatComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Hpr(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	AP	array with A stored in packed format.

Hpr(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	AP	array with A stored in packed format.

Hpr(FillMode, Double, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr(FillMode uplo, double alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	AP	array with A stored in packed format.

Hpr(FillMode, Single, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>)

This function performs the Hermitian rank-1 update A = alpha * x * x^H + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr(FillMode uplo, float alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	AP	array with A stored in packed format.

Hpr2(FillMode, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>)

This function performs the packed Hermitian rank-2 update A = alpha * (x * y^H + y * x^H) + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr2(FillMode uplo, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	AP	array with A stored in packed format.

Hpr2(FillMode, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>)

This function performs the packed Hermitian rank-2 update A = alpha * (x * y^H + y * x^H) + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr2(FillMode uplo, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	AP	array with A stored in packed format.

Hpr2(FillMode, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>)

This function performs the packed Hermitian rank-2 update A = alpha * (x * y^H + y * x^H) + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr2(FillMode uplo, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	AP	array with A stored in packed format.

Hpr2(FillMode, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>)

This function performs the packed Hermitian rank-2 update A = alpha * (x * y^H + y * x^H) + A where A is a n*n Hermitian Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Hpr2(FillMode uplo, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	AP	array with A stored in packed format.

MatinvBatchedC(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Declaration

public void MatinvBatchedC(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> Ainv, int lda_inv, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n
CudaDeviceVariable<CUdeviceptr>	A
System.Int32	lda
CudaDeviceVariable<CUdeviceptr>	Ainv
System.Int32	lda_inv
CudaDeviceVariable<System.Int32>	INFO
System.Int32	batchSize

MatinvBatchedD(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Declaration

public void MatinvBatchedD(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> Ainv, int lda_inv, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n
CudaDeviceVariable<CUdeviceptr>	A
System.Int32	lda
CudaDeviceVariable<CUdeviceptr>	Ainv
System.Int32	lda_inv
CudaDeviceVariable<System.Int32>	INFO
System.Int32	batchSize

MatinvBatchedS(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Declaration

public void MatinvBatchedS(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> Ainv, int lda_inv, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n
CudaDeviceVariable<CUdeviceptr>	A
System.Int32	lda
CudaDeviceVariable<CUdeviceptr>	Ainv
System.Int32	lda_inv
CudaDeviceVariable<System.Int32>	INFO
System.Int32	batchSize

MatinvBatchedZ(Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<Int32>, Int32)

Declaration

public void MatinvBatchedZ(int n, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> Ainv, int lda_inv, CudaDeviceVariable<int> INFO, int batchSize)

Parameters

Type	Name	Description
System.Int32	n
CudaDeviceVariable<CUdeviceptr>	A
System.Int32	lda
CudaDeviceVariable<CUdeviceptr>	Ainv
System.Int32	lda_inv
CudaDeviceVariable<System.Int32>	INFO
System.Int32	batchSize

Max(CudaDeviceVariable<cuDoubleComplex>, Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Max(CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Max(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Max(CudaDeviceVariable<cuDoubleComplex>, Int32, ref Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<cuDoubleComplex> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
System.Int32	result

Max(CudaDeviceVariable<cuFloatComplex>, Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Max(CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Max(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Max(CudaDeviceVariable<cuFloatComplex>, Int32, ref Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<cuFloatComplex> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
System.Int32	result

Max(CudaDeviceVariable<Double>, Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Max(CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Max(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Max(CudaDeviceVariable<Double>, Int32, ref Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<double> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
System.Int32	result

Max(CudaDeviceVariable<Single>, Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Max(CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Max(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Max(CudaDeviceVariable<Single>, Int32, ref Int32)

This function finds the (smallest) index of the element of the maximum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Max(CudaDeviceVariable<float> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
System.Int32	result

Min(CudaDeviceVariable<cuDoubleComplex>, Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Min(CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Min(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Min(CudaDeviceVariable<cuDoubleComplex>, Int32, ref Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<cuDoubleComplex> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
System.Int32	result

Min(CudaDeviceVariable<cuFloatComplex>, Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Min(CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Min(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Min(CudaDeviceVariable<cuFloatComplex>, Int32, ref Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<cuFloatComplex> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
System.Int32	result

Min(CudaDeviceVariable<Double>, Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Min(CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Min(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Min(CudaDeviceVariable<Double>, Int32, ref Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<double> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
System.Int32	result

Min(CudaDeviceVariable<Single>, Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public int Min(CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx

Returns

Type	Description
System.Int32

Min(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Int32>)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<int> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Int32>	result

Min(CudaDeviceVariable<Single>, Int32, ref Int32)

This function finds the (smallest) index of the element of the minimum magnitude.

First index starts at 1 (Fortran notation)

Declaration

public void Min(CudaDeviceVariable<float> x, int incx, ref int result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
System.Int32	result

Norm2(CudaDeviceVariable<cuDoubleComplex>, Int32)

This function computes the Euclidean norm of the vector x.

Declaration

public double Norm2(CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Returns

Type	Description
System.Double

Norm2(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Double>)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<double> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	result

Norm2(CudaDeviceVariable<cuDoubleComplex>, Int32, ref Double)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<cuDoubleComplex> x, int incx, ref double result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
System.Double	result

Norm2(CudaDeviceVariable<cuFloatComplex>, Int32)

This function computes the Euclidean norm of the vector x.

Declaration

public float Norm2(CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Returns

Type	Description
System.Single

Norm2(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Single>)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<float> result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	result

Norm2(CudaDeviceVariable<cuFloatComplex>, Int32, ref Single)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<cuFloatComplex> x, int incx, ref float result)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
System.Single	result

Norm2(CudaDeviceVariable<Double>, Int32)

This function computes the Euclidean norm of the vector x.

Declaration

public double Norm2(CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx

Returns

Type	Description
System.Double

Norm2(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	result

Norm2(CudaDeviceVariable<Double>, Int32, ref Double)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<double> x, int incx, ref double result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
System.Double	result

Norm2(CudaDeviceVariable<Single>, Int32)

This function computes the Euclidean norm of the vector x.

Declaration

public float Norm2(CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx

Returns

Type	Description
System.Single

Norm2(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	result

Norm2(CudaDeviceVariable<Single>, Int32, ref Single)

This function computes the Euclidean norm of the vector x.

Declaration

public void Norm2(CudaDeviceVariable<float> x, int incx, ref float result)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
System.Single	result

Rot(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<double> c, CudaDeviceVariable<cuDoubleComplex> s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
CudaDeviceVariable<System.Double>	c	Cosine component
CudaDeviceVariable<cuDoubleComplex>	s	Sine component

Rot(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<double> c, CudaDeviceVariable<double> s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
CudaDeviceVariable<System.Double>	c	Cosine component
CudaDeviceVariable<System.Double>	s	Sine component

Rot(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, Double, cuDoubleComplex)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, double c, cuDoubleComplex s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
System.Double	c	Cosine component
cuDoubleComplex	s	Sine component

Rot(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, Double, Double)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, double c, double s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy
System.Double	c	Cosine component
System.Double	s	Sine component

Rot(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<float> c, CudaDeviceVariable<cuFloatComplex> s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
CudaDeviceVariable<System.Single>	c	Cosine component
CudaDeviceVariable<cuFloatComplex>	s	Sine component

Rot(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<float> c, CudaDeviceVariable<float> s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
CudaDeviceVariable<System.Single>	c	Cosine component
CudaDeviceVariable<System.Single>	s	Sine component

Rot(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, Single, cuFloatComplex)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, float c, cuFloatComplex s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
System.Single	c	Cosine component
cuFloatComplex	s	Sine component

Rot(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, Single, Single)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, float c, float s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy
System.Single	c	Cosine component
System.Single	s	Sine component

Rot(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> c, CudaDeviceVariable<double> s)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy
CudaDeviceVariable<System.Double>	c	Cosine component
CudaDeviceVariable<System.Double>	s	Sine component

Rot(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, Double)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, double c, double s)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy
System.Double	c	Cosine component
System.Double	s	Sine component

Rot(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> c, CudaDeviceVariable<float> s)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy
CudaDeviceVariable<System.Single>	c	Cosine component
CudaDeviceVariable<System.Single>	s	Sine component

Rot(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, Single)

This function applies Givens rotation matrix G = |c s; -s c| to vectors x and y.

Declaration

public void Rot(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, float c, float s)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy
System.Single	c	Cosine component
System.Single	s	Sine component

Rotg(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(CudaDeviceVariable<cuDoubleComplex> a, CudaDeviceVariable<cuDoubleComplex> b, CudaDeviceVariable<cuDoubleComplex> c, CudaDeviceVariable<cuDoubleComplex> s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	a
CudaDeviceVariable<cuDoubleComplex>	b
CudaDeviceVariable<cuDoubleComplex>	c	Cosine component
CudaDeviceVariable<cuDoubleComplex>	s	Sine component

Rotg(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(CudaDeviceVariable<cuFloatComplex> a, CudaDeviceVariable<cuFloatComplex> b, CudaDeviceVariable<float> c, CudaDeviceVariable<cuFloatComplex> s)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	a
CudaDeviceVariable<cuFloatComplex>	b
CudaDeviceVariable<System.Single>	c	Cosine component
CudaDeviceVariable<cuFloatComplex>	s	Sine component

Rotg(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(CudaDeviceVariable<double> a, CudaDeviceVariable<double> b, CudaDeviceVariable<double> c, CudaDeviceVariable<double> s)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	a
CudaDeviceVariable<System.Double>	b
CudaDeviceVariable<System.Double>	c	Cosine component
CudaDeviceVariable<System.Double>	s	Sine component

Rotg(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(CudaDeviceVariable<float> a, CudaDeviceVariable<float> b, CudaDeviceVariable<float> c, CudaDeviceVariable<float> s)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	a
CudaDeviceVariable<System.Single>	b
CudaDeviceVariable<System.Single>	c	Cosine component
CudaDeviceVariable<System.Single>	s	Sine component

Rotg(ref cuDoubleComplex, ref cuDoubleComplex, ref Double, ref cuDoubleComplex)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(ref cuDoubleComplex a, ref cuDoubleComplex b, ref double c, ref cuDoubleComplex s)

Parameters

Type	Name	Description
cuDoubleComplex	a
cuDoubleComplex	b
System.Double	c	Cosine component
cuDoubleComplex	s	Sine component

Rotg(ref cuFloatComplex, ref cuFloatComplex, ref Single, ref cuFloatComplex)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(ref cuFloatComplex a, ref cuFloatComplex b, ref float c, ref cuFloatComplex s)

Parameters

Type	Name	Description
cuFloatComplex	a
cuFloatComplex	b
System.Single	c	Cosine component
cuFloatComplex	s	Sine component

Rotg(ref Double, ref Double, ref Double, ref Double)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(ref double a, ref double b, ref double c, ref double s)

Parameters

Type	Name	Description
System.Double	a
System.Double	b
System.Double	c	Cosine component
System.Double	s	Sine component

Rotg(ref Single, ref Single, ref Single, ref Single)

This function constructs the Givens rotation matrix G = |c s; -s c| that zeros out the second entry of a 2x1 vector (a; b)T

Declaration

public void Rotg(ref float a, ref float b, ref float c, ref float s)

Parameters

Type	Name	Description
System.Single	a
System.Single	b
System.Single	c	Cosine component
System.Single	s	Sine component

Rotm(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>)

This function constructs the modified Givens transformation H = |h11 h12; h21 h22| that zeros out the second entry of a 2x1 vector [sqrt(d1)*x1; sqrt(d2)*y1].

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(CudaDeviceVariable<double> d1, CudaDeviceVariable<double> d2, CudaDeviceVariable<double> x1, CudaDeviceVariable<double> y1, CudaDeviceVariable<double> param)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	d1
CudaDeviceVariable<System.Double>	d2
CudaDeviceVariable<System.Double>	x1
CudaDeviceVariable<System.Double>	y1
CudaDeviceVariable<System.Double>	param

Rotm(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function applies the modified Givens transformation H = |h11 h12; h21 h22| to vectors x and y.

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> param)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy
CudaDeviceVariable<System.Double>	param

Rotm(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double[])

This function applies the modified Givens transformation H = |h11 h12; h21 h22| to vectors x and y.

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, double[] param)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy
System.Double[]	param

Rotm(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>)

This function constructs the modified Givens transformation H = |h11 h12; h21 h22| that zeros out the second entry of a 2x1 vector [sqrt(d1)*x1; sqrt(d2)*y1].

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(CudaDeviceVariable<float> d1, CudaDeviceVariable<float> d2, CudaDeviceVariable<float> x1, CudaDeviceVariable<float> y1, CudaDeviceVariable<float> param)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	d1
CudaDeviceVariable<System.Single>	d2
CudaDeviceVariable<System.Single>	x1
CudaDeviceVariable<System.Single>	y1
CudaDeviceVariable<System.Single>	param

Rotm(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function applies the modified Givens transformation H = |h11 h12; h21 h22| to vectors x and y.

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> param)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy
CudaDeviceVariable<System.Single>	param

Rotm(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single[])

This function applies the modified Givens transformation H = |h11 h12; h21 h22| to vectors x and y.

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, float[] param)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy
System.Single[]	param

Rotm(ref Double, ref Double, ref Double, Double, Double[])

This function constructs the modified Givens transformation H = |h11 h12; h21 h22| that zeros out the second entry of a 2x1 vector [sqrt(d1)*x1; sqrt(d2)*y1].

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(ref double d1, ref double d2, ref double x1, double y1, double[] param)

Parameters

Type	Name	Description
System.Double	d1
System.Double	d2
System.Double	x1
System.Double	y1
System.Double[]	param

Rotm(ref Single, ref Single, ref Single, Single, Single[])

This function constructs the modified Givens transformation H = |h11 h12; h21 h22| that zeros out the second entry of a 2x1 vector [sqrt(d1)*x1; sqrt(d2)*y1].

The elements h11, h21, h12 and h22 of 2x2 matrix H are stored in param[1], param[2], param[3] and param[4], respectively.

The flag = param[0] defines the following predefined values for the matrix H entries:

flag=-1.0: H = |h11 h12; h21 h22|

flag= 0.0: H = |1.0 h12; h21 1.0|

flag= 1.0: H = |h11 1.0; -1.0 h22|

flag=-2.0: H = |1.0 0.0; 0.0 1.0|

Notice that the values -1.0, 0.0 and 1.0 implied by the flag are not stored in param.

Declaration

public void Rotm(ref float d1, ref float d2, ref float x1, float y1, float[] param)

Parameters

Type	Name	Description
System.Single	d1
System.Single	d2
System.Single	x1
System.Single	y1
System.Single[]	param

Sbmv(FillMode, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Sbmv(FillMode uplo, int k, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Sbmv(FillMode, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Sbmv(FillMode uplo, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Sbmv(FillMode, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Sbmv(FillMode uplo, int k, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, double beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Double	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Sbmv(FillMode, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric banded matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix with k subdiagonals and superdiagonals, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Sbmv(FillMode uplo, int k, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, float beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	k	number of sub- and super-diagonals of matrix A.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Single	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Scale(CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	alpha
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Scale(CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	alpha
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Scale(CudaDeviceVariable<Double>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(CudaDeviceVariable<double> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	alpha
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Scale(CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	alpha
CudaDeviceVariable<System.Double>	x
System.Int32	incx

Scale(CudaDeviceVariable<Single>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(CudaDeviceVariable<float> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	alpha
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Scale(CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	alpha
CudaDeviceVariable<System.Single>	x
System.Int32	incx

Scale(cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
cuDoubleComplex	alpha
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Scale(cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
cuFloatComplex	alpha
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Scale(Double, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(double alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
System.Double	alpha
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx

Scale(Double, CudaDeviceVariable<Double>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(double alpha, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
System.Double	alpha
CudaDeviceVariable<System.Double>	x
System.Int32	incx

Scale(Single, CudaDeviceVariable<cuFloatComplex>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(float alpha, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
System.Single	alpha
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx

Scale(Single, CudaDeviceVariable<Single>, Int32)

This function scales the vector x by the scalar and overwrites it with the result.

Declaration

public void Scale(float alpha, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
System.Single	alpha
CudaDeviceVariable<System.Single>	x
System.Int32	incx

SetMatrix<T>(Int32, Int32, T[], Int32, CudaDeviceVariable<T>, Int32)

copies a tile of rows x cols elements from a matrix hostSource in CPU memory space to a matrix devDest in GPU memory space. Both matrices are assumed to be stored in column major format, with the leading dimension (i.e. number of rows) of source matrix hostSource provided in ldHostSource, and the leading dimension of matrix devDest provided in ldDevDest.

Declaration

public static void SetMatrix<T>(int rows, int cols, T[] hostSource, int ldHostSource, CudaDeviceVariable<T> devDest, int ldDevDest)
    where T : struct

Parameters

Type	Name	Description
System.Int32	rows
System.Int32	cols
T[]	hostSource
System.Int32	ldHostSource
CudaDeviceVariable<T>	devDest
System.Int32	ldDevDest

Type Parameters

Name	Description
T

SetMatrixAsync<T>(Int32, Int32, T[], Int32, CudaDeviceVariable<T>, Int32, CUstream)

copies a tile of rows x cols elements from a matrix hostSource in CPU memory space to a matrix devDest in GPU memory space. Both matrices are assumed to be stored in column major format, with the leading dimension (i.e. number of rows) of source matrix hostSource provided in ldHostSource, and the leading dimension of matrix devDest provided in ldDevDest.

Declaration

public static void SetMatrixAsync<T>(int rows, int cols, T[] hostSource, int ldHostSource, CudaDeviceVariable<T> devDest, int ldDevDest, CUstream stream)
    where T : struct

Parameters

Type	Name	Description
System.Int32	rows
System.Int32	cols
T[]	hostSource
System.Int32	ldHostSource
CudaDeviceVariable<T>	devDest
System.Int32	ldDevDest
CUstream	stream

Type Parameters

Name	Description
T

SetVector<T>(T[], Int32, CudaDeviceVariable<T>, Int32)

copies elements from a vector hostSourceVector in CPU memory space to a vector devDestVector in GPU memory space. Storage spacing between consecutive elements is incrHostSource for the source vector hostSourceVector and incrDevDest for the destination vector devDestVector. Column major format for two-dimensional matrices is assumed throughout CUBLAS. Therefore, if the increment for a vector is equal to 1, this access a column vector while using an increment equal to the leading dimension of the respective matrix accesses a row vector.

Declaration

public static void SetVector<T>(T[] hostSourceVector, int incrHostSource, CudaDeviceVariable<T> devDestVector, int incrDevDest)
    where T : struct

Parameters

Type	Name	Description
T[]	hostSourceVector	Source vector in host memory
System.Int32	incrHostSource
CudaDeviceVariable<T>	devDestVector	Destination vector in device memory
System.Int32	incrDevDest

Type Parameters

Name	Description
T	Vector datatype

SetVectorAsync<T>(T[], Int32, CudaDeviceVariable<T>, Int32, CUstream)

copies elements from a vector hostSourceVector in CPU memory space to a vector devDestVector in GPU memory space. Storage spacing between consecutive elements is incrHostSource for the source vector hostSourceVector and incrDevDest for the destination vector devDestVector. Column major format for two-dimensional matrices is assumed throughout CUBLAS. Therefore, if the increment for a vector is equal to 1, this access a column vector while using an increment equal to the leading dimension of the respective matrix accesses a row vector.

Declaration

public static void SetVectorAsync<T>(T[] hostSourceVector, int incrHostSource, CudaDeviceVariable<T> devDestVector, int incrDevDest, CUstream stream)
    where T : struct

Parameters

Type	Name	Description
T[]	hostSourceVector	Source vector in host memory
System.Int32	incrHostSource
CudaDeviceVariable<T>	devDestVector	Destination vector in device memory
System.Int32	incrDevDest
CUstream	stream

Type Parameters

Name	Description
T	Vector datatype

Spmv(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Spmv(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> AP, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Spmv(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Spmv(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> AP, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Spmv(FillMode, Double, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Spmv(FillMode uplo, double alpha, CudaDeviceVariable<double> AP, CudaDeviceVariable<double> x, int incx, double beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Double	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Spmv(FillMode, Single, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric packed matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in packed format, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Spmv(FillMode uplo, float alpha, CudaDeviceVariable<float> AP, CudaDeviceVariable<float> x, int incx, float beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Single	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Spr(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Spr(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	AP	array with A stored in packed format.

Spr(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Spr(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	AP	array with A stored in packed format.

Spr(FillMode, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Spr(FillMode uplo, double alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	AP	array with A stored in packed format.

Spr(FillMode, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Spr(FillMode uplo, float alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	AP	array with A stored in packed format.

Spr2(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function performs the packed symmetric rank-2 update A = alpha * (x * y^T + y * x^T) + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Spr2(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	AP	array with A stored in packed format.

Spr2(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function performs the packed symmetric rank-2 update A = alpha * (x * y^T + y * x^T) + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Spr2(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	AP	array with A stored in packed format.

Spr2(FillMode, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>)

This function performs the packed symmetric rank-2 update A = alpha * (x * y^T + y * x^T) + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Spr2(FillMode uplo, double alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	AP	array with A stored in packed format.

Spr2(FillMode, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function performs the packed symmetric rank-2 update A = alpha * (x * y^T + y * x^T) + A where A is a n*n symmetric Matrix stored in packed format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Spr2(FillMode uplo, float alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	AP	array with A stored in packed format.

Stpttr(FillMode, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the conversion from the triangular packed format to the triangular format.

If uplo == CUBLAS_FILL_MODE_LOWER then the elements of AP are copied into the lower triangular part of the triangular matrix A and the upper part of A is left untouched.

If uplo == CUBLAS_FILL_MODE_UPPER then the elements of AP are copied into the upper triangular part of the triangular matrix A and the lower part of A is left untouched.

Declaration

public void Stpttr(FillMode uplo, int n, CudaDeviceVariable<float> AP, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix AP contains lower or upper part of matrix A.
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<System.Single>	AP	array with A stored in packed format.
CudaDeviceVariable<System.Single>	A	array of dimensions lda x n , with lda>=max(1,n). The opposite side of A is left untouched.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Strttp(FillMode, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>)

This function performs the conversion from the triangular format to the triangular packed format.

If uplo == CUBLAS_FILL_MODE_LOWER then the lower triangular part of the triangular matrix A is copied into the array AP.

If uplo == CUBLAS_FILL_MODE_UPPER then then the upper triangular part of the triangular matrix A is copied into the array AP

Declaration

public void Strttp(FillMode uplo, int n, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates which matrix A lower or upper part is referenced
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<System.Single>	A	array of dimensions lda x n , with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	AP	array with A stored in packed format.

Swap(CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleComplex>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleComplex>	y
System.Int32	incy

Swap(CudaDeviceVariable<cuDoubleReal>, Int32, CudaDeviceVariable<cuDoubleReal>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<cuDoubleReal> x, int incx, CudaDeviceVariable<cuDoubleReal> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuDoubleReal>	x
System.Int32	incx
CudaDeviceVariable<cuDoubleReal>	y
System.Int32	incy

Swap(CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatComplex>	x
System.Int32	incx
CudaDeviceVariable<cuFloatComplex>	y
System.Int32	incy

Swap(CudaDeviceVariable<cuFloatReal>, Int32, CudaDeviceVariable<cuFloatReal>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<cuFloatReal> x, int incx, CudaDeviceVariable<cuFloatReal> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<cuFloatReal>	x
System.Int32	incx
CudaDeviceVariable<cuFloatReal>	y
System.Int32	incy

Swap(CudaDeviceVariable<double1>, Int32, CudaDeviceVariable<double1>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<double1> x, int incx, CudaDeviceVariable<double1> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<double1>	x
System.Int32	incx
CudaDeviceVariable<double1>	y
System.Int32	incy

Swap(CudaDeviceVariable<float1>, Int32, CudaDeviceVariable<float1>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<float1> x, int incx, CudaDeviceVariable<float1> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<float1>	x
System.Int32	incx
CudaDeviceVariable<float1>	y
System.Int32	incy

Swap(CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Double>	x
System.Int32	incx
CudaDeviceVariable<System.Double>	y
System.Int32	incy

Swap(CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function interchanges the elements of vector x and y.

Declaration

public void Swap(CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
CudaDeviceVariable<System.Single>	x
System.Int32	incx
CudaDeviceVariable<System.Single>	y
System.Int32	incy

Symm(SideMode, FillMode, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symm(SideMode, FillMode, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symm(SideMode, FillMode, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symm(SideMode, FillMode, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symm(SideMode, FillMode, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuDoubleComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symm(SideMode, FillMode, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuFloatComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symm(SideMode, FillMode, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Double	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symm(SideMode, FillMode, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric matrix-matrix multiplication C = alphaAB + betaC if side==SideMode.Left or C = alphaBA + betaC if side==SideMode.Right where A is a symmetric matrix stored in lower or upper mode, B and C are m*n matrices, and alpha and beta are scalars.

Declaration

public void Symm(SideMode side, FillMode uplo, int m, int n, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of B.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Int32	m	number of rows of matrix C and B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix C and B, with matrix A sized accordingly.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Symv(FillMode, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha A * x + beta * y where A is a nn symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Symv(FillMode, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Symv(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha A * x + beta * y where A is a nn symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Symv(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Symv(FillMode, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuDoubleComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Symv(FillMode, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
cuFloatComplex	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Symv(FillMode, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx, double beta, CudaDeviceVariable<double> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Double	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Symv(FillMode, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric matrix-vector multiplication y = alpha * A * x + beta * y where A is a n*n symmetric matrix stored in lower or upper mode, x and y are vectors, and alpha and beta are scalars. n is given by x.Size.

Declaration

public void Symv(FillMode uplo, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx, float beta, CudaDeviceVariable<float> y, int incy)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
System.Single	beta	scalar used for multiplication, if beta==0 then y does not have to be a valid input.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.

Syr(FillMode, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr(FillMode, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr(FillMode, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr(FillMode, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr(FillMode, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, double alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr(FillMode, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-1 update A = alpha * x * x^T + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size.

Declaration

public void Syr(FillMode uplo, float alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> x, int incx, CudaDeviceVariable<cuDoubleComplex> y, int incy, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuDoubleComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> x, int incx, CudaDeviceVariable<cuFloatComplex> y, int incy, CudaDeviceVariable<cuFloatComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<cuFloatComplex>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, double alpha, CudaDeviceVariable<double> x, int incx, CudaDeviceVariable<double> y, int incy, CudaDeviceVariable<double> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Double>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2(FillMode, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-2 update A = alpha * (x * y^T + y * y^T) + A where A is a n*n symmetric Matrix stored in column-major format, x is a vector, and alpha is a scalar. n is given by x.Size = y.Size.

Declaration

public void Syr2(FillMode uplo, float alpha, CudaDeviceVariable<float> x, int incx, CudaDeviceVariable<float> y, int incy, CudaDeviceVariable<float> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.
CudaDeviceVariable<System.Single>	y	vector with n elements.
System.Int32	incy	stride between consecutive elements of y.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Syr2k(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syr2k(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syr2k(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syr2k(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syr2k(FillMode, Operation, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuDoubleComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syr2k(FillMode, Operation, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuFloatComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syr2k(FillMode, Operation, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Double	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syr2k(FillMode, Operation, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-k update C = alpha * (Op(A)Op(B)^T + Op(B)Op(A)^T) + beta * C where alpha and beta are scalars, and C is a symmetrux matrix stored in lower or upper mode, and A and B are matrices with dimensions Op(A) nk and Op(B) nk, respectively.

Declaration

public void Syr2k(FillMode uplo, Operation trans, int n, int k, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * k.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
cuDoubleComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
cuFloatComplex	beta	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, Double, CudaDeviceVariable<Double>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, double alpha, CudaDeviceVariable<double> A, int lda, double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Double	beta	scalar used for multiplication.
CudaDeviceVariable<System.Double>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrk(FillMode, Operation, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, Single, CudaDeviceVariable<Single>, Int32)

This function performs the symmetric rank-k update C = alpha * Op(A)Op(A)^T + beta * C where alpha and beta are scalars, and A, B and C are matrices stored in lower or upper mode, and A is a matrix with dimensions op(A) nk.

Declaration

public void Syrk(FillMode uplo, Operation trans, int n, int k, float alpha, CudaDeviceVariable<float> A, int lda, float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
System.Int32	n	number of columns of matrix op(B) and C.
System.Int32	k	number of columns of op(A) and rows of op(B).
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * k.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
System.Single	beta	scalar used for multiplication.
CudaDeviceVariable<System.Single>	C	array of dimensions ldb * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<System.Double>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<System.Single>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, ref cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, ref cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, ref cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, ref cuDoubleComplex beta, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
cuDoubleComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuDoubleComplex	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, ref cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, ref cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, ref cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, ref cuFloatComplex beta, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
cuFloatComplex	alpha	scalar used for multiplication.
CudaDeviceVariable<cuFloatComplex>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
cuFloatComplex	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, ref Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, ref Double, CudaDeviceVariable<Double>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, ref double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, ref double beta, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
System.Double	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Double>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Double	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<System.Double>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Syrkx(FillMode, Operation, Int32, Int32, ref Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, ref Single, CudaDeviceVariable<Single>, Int32)

This function performs a variation of the symmetric rank- update C = alpha * (Op(A)Op(B))^T + beta * C where alpha and beta are scalars, C is a symmetric matrix stored in lower or upper mode, and A and B are matrices with dimensions op(A) nk and op(B) n*k, respectively.

Declaration

public void Syrkx(FillMode uplo, Operation trans, int n, int k, ref float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, ref float beta, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix C lower or upper part, is stored, the other symmetric part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or transpose.
System.Int32	n	number of rows of matrix op(A), op(B) and C.
System.Int32	k	number of columns of matrix op(A) and op(B).
System.Single	alpha	scalar used for multiplication.
CudaDeviceVariable<System.Single>	A	array of dimension lda x k with lda>=max(1,n) if transa == CUBLAS_OP_N and lda x n with lda>=max(1,k) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb x k with ldb>=max(1,n) if transa == CUBLAS_OP_N and ldb x n with ldb>=max(1,k) otherwise.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
System.Single	beta	scalar used for multiplication, if beta==0, then C does not have to be a valid input.
CudaDeviceVariable<System.Single>	C	array of dimensions ldc x n with ldc>=max(1,n).
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Tbmv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the triangular banded matrix-vector multiplication x= Op(A) x where A is a triangular banded matrix, and x is a vector. n is given by x.Size.

Declaration

public void Tbmv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tbmv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the triangular banded matrix-vector multiplication x= Op(A) x where A is a triangular banded matrix, and x is a vector. n is given by x.Size.

Declaration

public void Tbmv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tbmv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the triangular banded matrix-vector multiplication x= Op(A) x where A is a triangular banded matrix, and x is a vector. n is given by x.Size.

Declaration

public void Tbmv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tbmv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the triangular banded matrix-vector multiplication x= Op(A) x where A is a triangular banded matrix, and x is a vector. n is given by x.Size.

Declaration

public void Tbmv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tbsv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function solves the triangular banded linear system with a single right-hand-side Op(A) x = b where A is a triangular banded matrix, and x and b is a vector. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tbsv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tbsv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function solves the triangular banded linear system with a single right-hand-side Op(A) x = b where A is a triangular banded matrix, and x and b is a vector. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tbsv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tbsv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function solves the triangular banded linear system with a single right-hand-side Op(A) x = b where A is a triangular banded matrix, and x and b is a vector. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tbsv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tbsv(FillMode, Operation, DiagType, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function solves the triangular banded linear system with a single right-hand-side Op(A) x = b where A is a triangular banded matrix, and x and b is a vector. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tbsv(FillMode uplo, Operation trans, DiagType diag, int k, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	k	number of sub- and super-diagonals of matrix A.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpmv(FillMode, Operation, DiagType, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the triangular packed matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in packed format, and x is a vector. n is given by x.Size.

Declaration

public void Tpmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuDoubleComplex> AP, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuDoubleComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpmv(FillMode, Operation, DiagType, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the triangular packed matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in packed format, and x is a vector. n is given by x.Size.

Declaration

public void Tpmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuFloatComplex> AP, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuFloatComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpmv(FillMode, Operation, DiagType, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function performs the triangular packed matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in packed format, and x is a vector. n is given by x.Size.

Declaration

public void Tpmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<double> AP, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Double>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpmv(FillMode, Operation, DiagType, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function performs the triangular packed matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in packed format, and x is a vector. n is given by x.Size.

Declaration

public void Tpmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<float> AP, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Single>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpsv(FillMode, Operation, DiagType, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function solves the packed triangular linear system with a single right-hand-side Op(A) x = b where A is a triangular matrix stored in packed format, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tpsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuDoubleComplex> AP, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuDoubleComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpsv(FillMode, Operation, DiagType, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32)

This function solves the packed triangular linear system with a single right-hand-side Op(A) x = b where A is a triangular matrix stored in packed format, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tpsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuFloatComplex> AP, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuFloatComplex>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpsv(FillMode, Operation, DiagType, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32)

This function solves the packed triangular linear system with a single right-hand-side Op(A) x = b where A is a triangular matrix stored in packed format, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tpsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<double> AP, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Double>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Tpsv(FillMode, Operation, DiagType, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32)

This function solves the packed triangular linear system with a single right-hand-side Op(A) x = b where A is a triangular matrix stored in packed format, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size. No test for singularity or near-singularity is included in this function.

Declaration

public void Tpsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<float> AP, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Single>	AP	array of dimensions lda * n, with lda >= max(1,n).
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trmv(FillMode, Operation, DiagType, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the triangular matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x is a vector. n is given by x.Size.

Declaration

public void Trmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trmv(FillMode, Operation, DiagType, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the triangular matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x is a vector. n is given by x.Size.

Declaration

public void Trmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trmv(FillMode, Operation, DiagType, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the triangular matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x is a vector. n is given by x.Size.

Declaration

public void Trmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trmv(FillMode, Operation, DiagType, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the triangular matrix-vector multiplication x= Op(A) x where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x is a vector. n is given by x.Size.

Declaration

public void Trmv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<double> alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<float> alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
cuDoubleComplex	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, cuDoubleComplex, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, cuDoubleComplex alpha, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> B, int ldb, CudaDeviceVariable<cuDoubleComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
cuDoubleComplex	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuDoubleComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
cuFloatComplex	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, cuFloatComplex, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, cuFloatComplex alpha, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> B, int ldb, CudaDeviceVariable<cuFloatComplex> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
cuFloatComplex	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<cuFloatComplex>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
System.Double	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, Double, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, double alpha, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> B, int ldb, CudaDeviceVariable<double> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
System.Double	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Double>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function solves the triangular linear system with multiple right-hand-sides Op(A)X = alphaB side==SideMode.Left or XOp(A) = alphaB if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the maindiagonal, X and B are m*n matrices, and alpha is a scalar.

The solution X overwrites the right-hand-sides B on exit.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
System.Single	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.

Trsm(SideMode, FillMode, Operation, DiagType, Int32, Int32, Single, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function performs the triangular matrix-matrix multiplication C = alphaOp(A) * B if side==SideMode.Left or C = alphaB * Op(A) if side==SideMode.Right where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, B and C are m*n matrices, and alpha is a scalar.

Notice that in order to achieve better parallelism CUBLAS differs from the BLAS API only for this routine. The BLAS API assumes an in-place implementation (with results written back to B), while the CUBLAS API assumes an out-of-place implementation (with results written into C). The application can obtain the in-place functionality of BLAS in the CUBLAS API by passing the address of the matrix B in place of the matrix C. No other overlapping in the input parameters is supported.

Declaration

public void Trsm(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, float alpha, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> B, int ldb, CudaDeviceVariable<float> C, int ldc)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A is on the left or right of X.
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
System.Int32	m	number of rows of matrix B, with matrix A sized accordingly.
System.Int32	n	number of columns of matrix B, with matrix A sized accordingly.
System.Single	alpha	scalar used for multiplication. If alpha==0 then A is not referenced and B does not have to be a valid input.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * m.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	B	array of dimensions ldb * n.
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B.
CudaDeviceVariable<System.Single>	C	array of dimensions ldc * n.
System.Int32	ldc	leading dimension of two-dimensional array used to store matrix C.

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<cuDoubleComplex> alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
CudaDeviceVariable<cuDoubleComplex>	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<cuFloatComplex>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<cuFloatComplex> alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
CudaDeviceVariable<cuFloatComplex>	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<Double>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<double> alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
CudaDeviceVariable<System.Double>	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, CudaDeviceVariable<Single>, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, CudaDeviceVariable<float> alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
CudaDeviceVariable<System.Single>	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, ref cuDoubleComplex, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, ref cuDoubleComplex alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
cuDoubleComplex	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, ref cuFloatComplex, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, ref cuFloatComplex alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
cuFloatComplex	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, ref Double, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, ref double alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
System.Double	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

TrsmBatched(SideMode, FillMode, Operation, DiagType, Int32, Int32, ref Single, CudaDeviceVariable<CUdeviceptr>, Int32, CudaDeviceVariable<CUdeviceptr>, Int32, Int32)

This function solves an array of triangular linear systems with multiple right-hand-sides.

The solution overwrites the right-hand-sides on exit.

No test for singularity or near-singularity is included in this function.

This function is intended to be used for matrices of small sizes where the launch overhead is a significant factor. The current implementation limits the dimensions m and n to 32.

Declaration

public void TrsmBatched(SideMode side, FillMode uplo, Operation trans, DiagType diag, int m, int n, ref float alpha, CudaDeviceVariable<CUdeviceptr> A, int lda, CudaDeviceVariable<CUdeviceptr> B, int ldb, int batchCount)

Parameters

Type	Name	Description
SideMode	side	indicates if matrix A[i] is on the left or right of X[i].
FillMode	uplo	indicates if matrix A[i] lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A[i]) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A[i] are unity and should not be accessed.
System.Int32	m	number of rows of matrix B[i], with matrix A[i] sized accordingly.
System.Int32	n	number of columns of matrix B[i], with matrix A[i] is sized accordingly.
System.Single	alpha	scalar used for multiplication, if alpha==0 then A[i] is not referenced and B[i] does not have to be a valid input.
CudaDeviceVariable<CUdeviceptr>	A	array of device pointers with each array/device pointerarray of dim. lda x m with lda>=max(1,m) if transa==CUBLAS_OP_N and lda x n with lda>=max(1,n) otherwise.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A[i].
CudaDeviceVariable<CUdeviceptr>	B	array of device pointers with each array/device pointerarrayof dim. ldb x n with ldb>=max(1,m)
System.Int32	ldb	leading dimension of two-dimensional array used to store matrix B[i].
System.Int32	batchCount

Trsv(FillMode, Operation, DiagType, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function solves the triangular linear system with a single right-hand-side Op(A)x = b where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size.

Declaration

public void Trsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trsv(FillMode, Operation, DiagType, CudaDeviceVariable<cuFloatComplex>, Int32, CudaDeviceVariable<cuFloatComplex>, Int32)

This function solves the triangular linear system with a single right-hand-side Op(A)x = b where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size.

Declaration

public void Trsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<cuFloatComplex> A, int lda, CudaDeviceVariable<cuFloatComplex> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<cuFloatComplex>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuFloatComplex>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trsv(FillMode, Operation, DiagType, CudaDeviceVariable<Double>, Int32, CudaDeviceVariable<Double>, Int32)

This function solves the triangular linear system with a single right-hand-side Op(A)x = b where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size.

Declaration

public void Trsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<double> A, int lda, CudaDeviceVariable<double> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Double>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Double>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Trsv(FillMode, Operation, DiagType, CudaDeviceVariable<Single>, Int32, CudaDeviceVariable<Single>, Int32)

This function solves the triangular linear system with a single right-hand-side Op(A)x = b where A is a triangular matrix stored in lower or upper mode with or without the main diagonal, and x and b are vectors. The solution x overwrites the right-hand-sides b on exit. n is given by x.Size.

Declaration

public void Trsv(FillMode uplo, Operation trans, DiagType diag, CudaDeviceVariable<float> A, int lda, CudaDeviceVariable<float> x, int incx)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix A lower or upper part is stored, the other part is not referenced and is inferred from the stored elements.
Operation	trans	operation op(A) that is non- or (conj.) transpose.
DiagType	diag	indicates if the elements on the main diagonal of matrix A are unity and should not be accessed.
CudaDeviceVariable<System.Single>	A	array of dimensions lda * n, with lda >= max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<System.Single>	x	vector with n elements.
System.Int32	incx	stride between consecutive elements of x.

Ztpttr(FillMode, Int32, CudaDeviceVariable<cuDoubleComplex>, CudaDeviceVariable<cuDoubleComplex>, Int32)

This function performs the conversion from the triangular packed format to the triangular format.

If uplo == CUBLAS_FILL_MODE_LOWER then the elements of AP are copied into the lower triangular part of the triangular matrix A and the upper part of A is left untouched.

If uplo == CUBLAS_FILL_MODE_UPPER then the elements of AP are copied into the upper triangular part of the triangular matrix A and the lower part of A is left untouched.

Declaration

public void Ztpttr(FillMode uplo, int n, CudaDeviceVariable<cuDoubleComplex> AP, CudaDeviceVariable<cuDoubleComplex> A, int lda)

Parameters

Type	Name	Description
FillMode	uplo	indicates if matrix AP contains lower or upper part of matrix A.
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	AP	array with A stored in packed format.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda x n , with lda>=max(1,n). The opposite side of A is left untouched.
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.

Ztrttp(FillMode, Int32, CudaDeviceVariable<cuDoubleComplex>, Int32, CudaDeviceVariable<cuDoubleComplex>)

This function performs the conversion from the triangular format to the triangular packed format.

If uplo == CUBLAS_FILL_MODE_LOWER then the lower triangular part of the triangular matrix A is copied into the array AP.

If uplo == CUBLAS_FILL_MODE_UPPER then then the upper triangular part of the triangular matrix A is copied into the array AP

Declaration

public void Ztrttp(FillMode uplo, int n, CudaDeviceVariable<cuDoubleComplex> A, int lda, CudaDeviceVariable<cuDoubleComplex> AP)

Parameters

Type	Name	Description
FillMode	uplo	indicates which matrix A lower or upper part is referenced
System.Int32	n	number of rows and columns of matrix A.
CudaDeviceVariable<cuDoubleComplex>	A	array of dimensions lda x n , with lda>=max(1,n).
System.Int32	lda	leading dimension of two-dimensional array used to store matrix A.
CudaDeviceVariable<cuDoubleComplex>	AP	array with A stored in packed format.