Functions related to sparse matrices and sparse vectors are defined. More...

Classes
class	SpVec
	Sparse vector class. More...

Functions
template<typename T >
void	SpmvCsr (T val, int cind, int rptr, T x, T *y, int N)
	Perform SpMV using the CSR format. More...

template<typename T >
void	SpmvCsr (T val, int cind, int rptr, T diag, T x, T y, int N)
	Perform SpMV using the CSR format, which stores diagonal elements separately. More...

template<typename T , int bnl, int bnw>
void	SpmvBcsr (T bval, int bcind, int brptr, T x, T *y, int N)
	Perform SpMV using the BCSR format. More...

template<typename T >
void	SpmvSell (T val, int cind, int wid, int len, T x, T *y, int N)
	Perform SpMV using the sliced-ELLPACK (SELL-c) format. More...

template<typename T >
void	SptrsvCsr_l (T val, int cind, int rptr, T x, T *y, int N)
	Perform the sparse lower triangular solve on a matrix stored in the CSR format. More...

template<typename T >
void	SptrsvCsr_u (T val, int cind, int rptr, T x, T *y, int N)
	Perform the sparse upper triangular solve on a matrix stored in the CSR format. More...

template<typename T >
void	SptrsvCsr_l (T val, int cind, int rptr, T x, T y, int N, int cptr, int cnum)
	Perform the sparse lower triangular solve in parallel on a AMC ordered matrix stored in the CSR format. More...

template<typename T >
void	SptrsvCsr_u (T val, int cind, int rptr, T x, T y, int N, int cptr, int cnum)
	Perform the sparse upper triangular solve in parallel on a AMC ordered matrix stored in the CSR format. More...

template<typename T >
void	SptrsvCsr_l (T val, int cind, int rptr, T x, T y, int N, int cptr, int cnum, int bsize)
	Perform the sparse lower triangular solve in parallel on a ABMC ordered matrix stored in the CSR format. More...

template<typename T >
void	SptrsvCsr_u (T val, int cind, int rptr, T x, T y, int N, int cptr, int cnum, int bsize)
	Perform the sparse upper triangular solve in parallel on a ABMC ordered matrix stored in the CSR format. More...

template<typename T >
void	SptrsvCsr_l (T val, int cind, int rptr, T x, T *y, int N, int bnum)
	Perform the block Jacobi sparse lower triangular solve on a matrix stored in the CSR format. More...

template<typename T >
void	SptrsvCsr_u (T val, int cind, int rptr, T x, T *y, int N, int bnum)
	Perform the block Jacobi sparse upper triangular solve on a matrix stored in the CSR format. More...

template<typename T , int bnl, int bnw>
void	SptrsvBcsr_l (T bval, int bcind, int brptr, T x, T *y, int N)
	Perform the sparse lower triangular solve for a matrix stored in the BCSR format. More...

template<typename T , int bnl, int bnw>
void	SptrsvBcsr_u (T bval, int bcind, int brptr, T x, T *y, int N)
	Perform the sparse upper triangular solve for a matrix stored in the CSR format. More...

template<typename T , int bnl, int bnw>
void	SptrsvBcsr_l (T bval, int bcind, int brptr, T x, T y, int N, int cptr, int cnum, int bsize)
	Perform the sparse lower triangular solve for a ABMC reordered matrix stored in the BCSR format. More...

template<typename T , int bnl, int bnw>
void	SptrsvBcsr_u (T bval, int bcind, int brptr, T x, T y, int N, int cptr, int cnum, int bsize)
	Perform the sparse upper triangular solve for a ABMC reordered matrix stored in the CSR format. More...

Detailed Description

Functions related to sparse matrices and sparse vectors are defined.

Function Documentation

◆ SpmvBcsr()

template<typename T , int bnl, int bnw>

void senk::sparse::SpmvBcsr	(	T *	bval,
		int *	bcind,
		int *	brptr,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform SpMV using the BCSR format.

Template Parameters

T	The Type of the matrix and the vectors.
bnl	The number of rows of the block.
bnw	The number of columns of the block.

Parameters

bval	A val array in the BCSR format.
bcind	A col-index array in the BCSR format.
brptr	A row-pointer array in the BCSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 71 of file senk_sparse.hpp.

◆ SpmvCsr() [1/2]

template<typename T >

void senk::sparse::SpmvCsr	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	diag,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform SpMV using the CSR format, which stores diagonal elements separately.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
diag	An array that stores diagonal elements.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 48 of file senk_sparse.hpp.

◆ SpmvCsr() [2/2]

template<typename T >

void senk::sparse::SpmvCsr	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform SpMV using the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 26 of file senk_sparse.hpp.

◆ SpmvSell()

template<typename T >

void senk::sparse::SpmvSell	(	T *	val,
		int *	cind,
		int *	wid,
		int	len,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform SpMV using the sliced-ELLPACK (SELL-c) format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the SELL-c format.
cind	A col-index array in the SELL-c format.
wid	An array that indicates the starting position of the slices.
len	The size of the slices.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 104 of file senk_sparse.hpp.

◆ SptrsvBcsr_l() [1/2]

template<typename T , int bnl, int bnw>

void senk::sparse::SptrsvBcsr_l	(	T *	bval,
		int *	bcind,
		int *	brptr,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform the sparse lower triangular solve for a matrix stored in the BCSR format.

Template Parameters

T	The Type of the matrix and the vectors.
bnl	The number of rows of the block.
bnw	The number of columns of the block.

Parameters

bval	A val array in the BCSR format.
bcind	A block col-index array in the BCSR format.
brptr	A block row-pointer array in the BCSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 393 of file senk_sparse.hpp.

◆ SptrsvBcsr_l() [2/2]

template<typename T , int bnl, int bnw>

void senk::sparse::SptrsvBcsr_l	(	T *	bval,
		int *	bcind,
		int *	brptr,
		T *	x,
		T *	y,
		int	N,
		int *	cptr,
		int	cnum,
		int	bsize
	)

inline

Perform the sparse lower triangular solve for a ABMC reordered matrix stored in the BCSR format.

Template Parameters

T	The Type of the matrix and the vectors.
bnl	The number of rows of the block.
bnw	The number of columns of the block.

Parameters

bval	A val array in the BCSR format.
bcind	A block col-index array in the BCSR format.
brptr	A block row-pointer array in the BCSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
cptr	The starting index of each color is stored.
cnum	The number of colors.
bsize	The number of rows/columns of the blocks used in ABMC.

Definition at line 482 of file senk_sparse.hpp.

◆ SptrsvBcsr_u() [1/2]

template<typename T , int bnl, int bnw>

void senk::sparse::SptrsvBcsr_u	(	T *	bval,
		int *	bcind,
		int *	brptr,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform the sparse upper triangular solve for a matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.
bnl	The number of rows of the block.
bnw	The number of columns of the block.

Parameters

bval	A val array in the BCSR format.
bcind	A block col-index array in the BCSR format.
brptr	A block row-pointer array in the BCSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 430 of file senk_sparse.hpp.

◆ SptrsvBcsr_u() [2/2]

template<typename T , int bnl, int bnw>

void senk::sparse::SptrsvBcsr_u	(	T *	bval,
		int *	bcind,
		int *	brptr,
		T *	x,
		T *	y,
		int	N,
		int *	cptr,
		int	cnum,
		int	bsize
	)

inline

Perform the sparse upper triangular solve for a ABMC reordered matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.
bnl	The number of rows of the block.
bnw	The number of columns of the block.

Parameters

bval	A val array in the BCSR format.
bcind	A block col-index array in the BCSR format.
brptr	A block row-pointer array in the BCSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
cptr	The starting index of each color is stored.
cnum	The number of colors.
bsize	The number of rows/columns of the blocks used in ABMC.

Definition at line 534 of file senk_sparse.hpp.

◆ SptrsvCsr_l() [1/4]

template<typename T >

void senk::sparse::SptrsvCsr_l	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform the sparse lower triangular solve on a matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 133 of file senk_sparse.hpp.

◆ SptrsvCsr_l() [2/4]

template<typename T >

void senk::sparse::SptrsvCsr_l	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N,
		int *	cptr,
		int	cnum
	)

inline

Perform the sparse lower triangular solve in parallel on a AMC ordered matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
cptr	The starting index of each color is stored.
cnum	The number of colors.

Definition at line 181 of file senk_sparse.hpp.

◆ SptrsvCsr_l() [3/4]

template<typename T >

void senk::sparse::SptrsvCsr_l	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N,
		int *	cptr,
		int	cnum,
		int	bsize
	)

inline

Perform the sparse lower triangular solve in parallel on a ABMC ordered matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
cptr	The starting index of each color is stored.
cnum	The number of colors.
bsize	The number of rows/columns of the blocks used in ABMC.

Definition at line 251 of file senk_sparse.hpp.

◆ SptrsvCsr_l() [4/4]

template<typename T >

void senk::sparse::SptrsvCsr_l	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N,
		int	bnum
	)

inline

Perform the block Jacobi sparse lower triangular solve on a matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
bptr	The starting index of each block is stored.
bnum	The number of the blocks.

Definition at line 329 of file senk_sparse.hpp.

◆ SptrsvCsr_u() [1/4]

template<typename T >

void senk::sparse::SptrsvCsr_u	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N
	)

inline

Perform the sparse upper triangular solve on a matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.

Definition at line 155 of file senk_sparse.hpp.

◆ SptrsvCsr_u() [2/4]

template<typename T >

void senk::sparse::SptrsvCsr_u	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N,
		int *	cptr,
		int	cnum
	)

inline

Perform the sparse upper triangular solve in parallel on a AMC ordered matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
cptr	The starting index of each color is stored.
cnum	The number of colors.

Definition at line 215 of file senk_sparse.hpp.

◆ SptrsvCsr_u() [3/4]

template<typename T >

void senk::sparse::SptrsvCsr_u	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N,
		int *	cptr,
		int	cnum,
		int	bsize
	)

inline

Perform the sparse upper triangular solve in parallel on a ABMC ordered matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
cptr	The starting index of each color is stored.
cnum	The number of colors.
bsize	The number of rows/columns of the blocks used in ABMC.

Definition at line 290 of file senk_sparse.hpp.

◆ SptrsvCsr_u() [4/4]

template<typename T >

void senk::sparse::SptrsvCsr_u	(	T *	val,
		int *	cind,
		int *	rptr,
		T *	x,
		T *	y,
		int	N,
		int	bnum
	)

inline

Perform the block Jacobi sparse upper triangular solve on a matrix stored in the CSR format.

Template Parameters

T	The Type of the matrix and the vectors.

Parameters

val	A val array in the CSR format.
cind	A col-index array in the CSR format.
rptr	A row-pointer array in the CSR format.
x	Input vector of size N.
y	Output vector of size N.
N	The size of vectors.
bptr	The starting index of each block is stored.
bnum	The number of the blocks.

Definition at line 360 of file senk_sparse.hpp.

Classes

Functions

Detailed Description

Function Documentation

◆ SpmvBcsr()

◆ SpmvCsr() [1/2]

◆ SpmvCsr() [2/2]

◆ SpmvSell()

◆ SptrsvBcsr_l() [1/2]

◆ SptrsvBcsr_l() [2/2]

◆ SptrsvBcsr_u() [1/2]

◆ SptrsvBcsr_u() [2/2]

◆ SptrsvCsr_l() [1/4]

◆ SptrsvCsr_l() [2/4]

◆ SptrsvCsr_l() [3/4]

◆ SptrsvCsr_l() [4/4]

◆ SptrsvCsr_u() [1/4]

◆ SptrsvCsr_u() [2/4]

◆ SptrsvCsr_u() [3/4]

◆ SptrsvCsr_u() [4/4]