#include <LAMAInterface.hpp>

Inheritance diagram for lama::BLAS1Interface< T >:

Public Member Functions
	BLAS1Interface ()
	Default constructor, initializes variables with NULL.
Data Fields
void(*	scal )(const IndexType n, const T alpha, T x, const IndexType incX, SyncToken syncToken)
	scal replaces vector x with alpha * x.
T(*	nrm2 )(const IndexType n, const T x, const IndexType incX, SyncToken syncToken)
	nrm2 computes the Euclidean norm of the n-vector x (with storage increment incX).
T(*	asum )(const IndexType n, const T x, const IndexType incX, SyncToken syncToken)
	asum computes the sum of the absolute values of the elements of vector x
IndexType(*	iamax )(const IndexType n, const T x, const IndexType incX, SyncToken syncToken)
	max norm: maxnorm(x) = max( x_i )
T(*	viamax )(const IndexType n, const T x, const IndexType incX, SyncToken syncToken)
	iamax finds the smallest index of the maximum magnitude element of vector x
void(*	swap )(const IndexType n, T x, const IndexType incX, T y, const IndexType incY, SyncToken *syncToken)
	swap interchanges vector x with vector y.
void(*	copy )(const IndexType n, const T x, const IndexType incX, T y, const IndexType incY, SyncToken *syncToken)
	copy copies the vector x to the vector y.
void(*	axpy )(const IndexType n, T alpha, const T x, const IndexType incX, T y, const IndexType incY, SyncToken *syncToken)
	axpy multiplies vector x by scalar alpha and adds the result to vector y.
T(*	dot )(const IndexType n, const T x, const IndexType incX, const T y, const IndexType inc, SyncToken *syncToken)
	dot computes the dot product of two vectors.
void(*	sum )(const IndexType n, T alpha, const T x, T beta, const T y, T z, SyncToken syncToken)
	TODO: need dotu ?!
void(*	rot )(const IndexType N, T X, const IndexType incX, T Y, const IndexType incY, const T c, const T s, SyncToken *syncToken)
	rot multiplies a 2×2 matrix with the 2×n matrix
void(*	rotm )(const IndexType N, T X, const IndexType incX, T Y, const IndexType incY, const T P, SyncToken syncToken)
	rotm ?
void(*	rotg )(T a, T b, T c, T s, SyncToken *syncToken)
	rotg The function cublasSrot(n, x, incX, y, incY, c, s) normally is called next to apply the transformation to a 2×n matrix
void(*	rotmg )(T d1, T d2, T b1, const T b2, T p, SyncToken *syncToken)
	Computes the parameters for a modified Givens rotation.
void(*	ass )(const IndexType n, const T value, T x, SyncToken syncToken)
	ass The function ass() assigns one scalar value to a vector of the given size.

template<typename T>
struct lama::BLAS1Interface< T >

Constructor & Destructor Documentation

template<typename T >

lama::BLAS1Interface< T >::BLAS1Interface ( )

Default constructor, initializes variables with NULL.

Field Documentation

template<typename T>

void(* lama::BLAS1Interface< T >::ass)(const IndexType n, const T value, T *x, SyncToken *syncToken)

ass The function ass() assigns one scalar value to a vector of the given size.

This function is OpenMP implementation of lama::BLAS1Interface::ass

The function ass() assigns one scalar value to a vector of the given size.

Parameters:

[in]	n	size of the vector
[in]	value	scalar value, which should be assign to the whole vector
[out]	x	vector, the values should be assigned to

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

T(* lama::BLAS1Interface< T >::asum)(const IndexType n, const T *x, const IndexType incX, SyncToken *syncToken)

asum computes the sum of the absolute values of the elements of vector x

asum(x) = sum_{i = 1}^{n}( x_i )

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x return the sum of absolute values (returns zero if n<=0 or incX <=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::axpy)(const IndexType n, T alpha, const T *x, const IndexType incX, T *y, const IndexType incY, SyncToken *syncToken)

axpy multiplies vector x by scalar alpha and adds the result to vector y.

y = alpha * y + beta * x

Parameters:

[in]	n	number of elements in input vectors.
[in]	alpha	scalar multiplier
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x
[in]	y	vector with n elements
[in]	incY	storage spacing between elements of y
[out]	y	result (unchanged if n<=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::copy)(const IndexType n, const T *x, const IndexType incX, T *y, const IndexType incY, SyncToken *syncToken)

copy copies the vector x to the vector y.

y = x

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x
[in]	y	vector with n elements
[in]	incY	storage spacing between elements of y
[out]	y	contains vector x

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

T(* lama::BLAS1Interface< T >::dot)(const IndexType n, const T *x, const IndexType incX, const T *y, const IndexType inc, SyncToken *syncToken)

dot computes the dot product of two vectors.

It returns the dot product of the vectors x and y if successfull, and 0.0 otherwise. adds the result to vector y.

dot = sum_{i = 1}^{n}( x_i * y_i )

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x
[in]	y	vector with n elements
[in]	incY	storage spacing between elements of y return dot product (returns zero if n <= 0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

IndexType(* lama::BLAS1Interface< T >::iamax)(const IndexType n, const T *x, const IndexType incX, SyncToken *syncToken)

max norm: maxnorm(x) = max( x_i )

iamax finds the smallest index of the maximum magnitude element of vector x

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x return the smallest index (returns zero if n <=0 or incX <=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

T(* lama::BLAS1Interface< T >::nrm2)(const IndexType n, const T *x, const IndexType incX, SyncToken *syncToken)

nrm2 computes the Euclidean norm of the n-vector x (with storage increment incX).

nrm2(x) = sqrt( sum_{i = 1}^{n}( x_i^2 ) )

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x return the Euclidian norm (returns zero if n <=0, incX <=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::rot)(const IndexType N, T *X, const IndexType incX, T *Y, const IndexType incY, const T c, const T s, SyncToken *syncToken)

rot multiplies a 2×2 matrix with the 2×n matrix

Parameters:

[in]	n	number of elements in input vectors.
[in,out]	x	vector with n elements
[in]	incX	storage spacing between elements of x
[in,out]	y	vector with n elements
[in]	incY	storage spacing between elements of y
[in]	c	element of rotation matrix
[in]	s	element of rotation matrix

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::rotg)(T *a, T *b, T *c, T *s, SyncToken *syncToken)

rotg The function cublasSrot(n, x, incX, y, incY, c, s) normally is called next to apply the transformation to a 2×n matrix

Parameters:

[in,out]	a	scalar
[in,out]	b	scalar
[in,out]	c	result
[in,out]	s	result

Referenced by lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::rotm)(const IndexType N, T *X, const IndexType incX, T *Y, const IndexType incY, const T *P, SyncToken *syncToken)

rotm ?

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x
[in]	y	vector with n elements
[in]	incY	storage spacing between elements of y
[in]	*P	? //in cublas sparam
[out]	x	rotated vector x(unchanged if n<=0)
[out]	y	rotated vector y(unchanged if n<=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::rotmg)(T *d1, T *d2, T *b1, const T b2, T *p, SyncToken *syncToken)

Computes the parameters for a modified Givens rotation.

Parameters:

[in]	d1	scaling factor for the x-coordinate of the input vector
[in]	d2	scaling factor for the y-coordinate of the input vector
[in]	b1	scalar
[in]	b2	scalar
[out]	d1	changed to represent the effect of the transformation
[out]	d2	changed to represent the effect of the transformation
[out]	b1	changed to represent the effect of the transformation
[out]	P	? //in cublas sparam

Referenced by lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::scal)(const IndexType n, const T alpha, T *x, const IndexType incX, SyncToken *syncToken)

scal replaces vector x with alpha * x.

x = alpha * x

Parameters:

[in]	n	number of elements in input vectors.
[in]	alpha	scalar multiplier
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x
[out]	x	vector x(unchanged if n<=0 or incX <=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::sum)(const IndexType n, T alpha, const T *x, T beta, const T *y, T *z, SyncToken *syncToken)

TODO: need dotu ?!

sum: z = alpha * x + beta * y

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

void(* lama::BLAS1Interface< T >::swap)(const IndexType n, T *x, const IndexType incX, T *y, const IndexType incY, SyncToken *syncToken)

swap interchanges vector x with vector y.

x <-> y

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x
[in]	y	vector with n elements
[in]	incY	storage spacing between elements of y
[out]	x	vector x(unchanged if n<=0)
[out]	y	vector y(unchanged if n<=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

template<typename T>

T(* lama::BLAS1Interface< T >::viamax)(const IndexType n, const T *x, const IndexType incX, SyncToken *syncToken)

iamax finds the smallest index of the maximum magnitude element of vector x

Parameters:

[in]	n	number of elements in input vectors.
[in]	x	vector with n elements
[in]	incX	storage spacing between elements of x return the smallest index (returns zero if n <=0 or incX <=0)

Referenced by lama::CUDAInterface::CUDAInterface(), and lama::OpenMPInterface::OpenMPInterface().

The documentation for this struct was generated from the following files:

/home/brandes/workspace/LAMA/src/lama/LAMAInterface.hpp
/home/brandes/workspace/LAMA/src/lama/LAMAInterface.cpp

Public Member Functions

Data Fields

template<typename T> struct lama::BLAS1Interface< T >

Constructor & Destructor Documentation

Field Documentation

template<typename T>
struct lama::BLAS1Interface< T >