mod_elpa_hermitian_multiply.F90 File Reference

#include "config-f90.h"
#include "../general/precision_macros.h"
#include "elpa_hermitian_multiply_template.F90"

Macros
#define	NVTX_RANGE_PUSH(x)
#define	NVTX_RANGE_POP(x)
#define	REALCASE   1
#define	DOUBLE_PRECISION
#define	REALCASE   1
#define	SINGLE_PRECISION
#define	COMPLEXCASE   1
#define	DOUBLE_PRECISION
#define	COMPLEXCASE   1
#define	SINGLE_PRECISION

Functions/Subroutines
program	__mod_elpa_hermitian_multiply_F90__
function	elpa_hermitian_multiply_a_h_a_real_double_impl (obj, uplo_a, uplo_c, ncb, a, b, ldb, ldbcols, c, ldc, ldccols)
	elpa_hermitian_multiply_a_h_a_real_double_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed
function	elpa_hermitian_multiply_d_ptr_real_double_impl (obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols)
	elpa_hermitian_multiply_d_ptr_real_double_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed
function	elpa_hermitian_multiply_a_h_a_real_single_impl (obj, uplo_a, uplo_c, ncb, a, b, ldb, ldbcols, c, ldc, ldccols)
	elpa_hermitian_multiply_real_a_h_a_single_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed
function	elpa_hermitian_multiply_d_ptr_real_single_impl (obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols)
	elpa_hermitian_multiply_real_d_ptr_single_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed
function	elpa_hermitian_multiply_a_h_a_complex_double_impl (obj, uplo_a, uplo_c, ncb, a, b, ldb, ldbcols, c, ldc, ldccols)
	elpa_hermitian_multiply_a_h_a_complex_double_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed
function	elpa_hermitian_multiply_d_ptr_complex_double_impl (obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols)
	elpa_hermitian_multiply_a_h_a_complex_double_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed
function	elpa_hermitian_multiply_a_h_a_complex_single_impl (obj, uplo_a, uplo_c, ncb, a, b, ldb, ldbcols, c, ldc, ldccols)
	elpa_hermitian_multiply_a_h_a_complex_single_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed
function	elpa_hermitian_multiply_d_ptr_complex_single_impl (obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols)
	elpa_hermitian_multiply_d_ptr_complex_single_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Macro Definition Documentation

COMPLEXCASE [1/2]

#define COMPLEXCASE   1

COMPLEXCASE [2/2]

#define COMPLEXCASE   1

DOUBLE_PRECISION [1/2]

#define DOUBLE_PRECISION

DOUBLE_PRECISION [2/2]

#define DOUBLE_PRECISION

NVTX_RANGE_POP

#define NVTX_RANGE_POP

(

x

)

NVTX_RANGE_PUSH

#define NVTX_RANGE_PUSH

(

x

)

REALCASE [1/2]

#define REALCASE   1

REALCASE [2/2]

#define REALCASE   1

SINGLE_PRECISION [1/2]

#define SINGLE_PRECISION

SINGLE_PRECISION [2/2]

#define SINGLE_PRECISION

Function/Subroutine Documentation

__mod_elpa_hermitian_multiply_F90__()

program __mod_elpa_hermitian_multiply_F90__

elpa_hermitian_multiply_a_h_a_complex_double_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_a_h_a_complex_double_impl ( obj, uplo_a, uplo_c, ncb, a, b, ldb, ldbcols, c, ldc, ldccols )

private

elpa_hermitian_multiply_a_h_a_complex_double_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a
obj%local_ncols	leading dimension of matrix a, set with class method objset("local_nrows",value)
ldaCols	columns of matrix a
b	matrix b
ldb	leading dimension of matrix b
ldbCols	columns of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c
ldc	leading dimension of matrix c

Returns

success

elpa_hermitian_multiply_a_h_a_complex_single_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_a_h_a_complex_single_impl ( obj, uplo_a, uplo_c, ncb, a, b, ldb, ldbcols, c, ldc, ldccols )

private

elpa_hermitian_multiply_a_h_a_complex_single_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a
obj%local_ncols	leading dimension of matrix a, set with class method objset("local_nrows",value)
ldaCols	columns of matrix a
b	matrix b
ldb	leading dimension of matrix b
ldbCols	columns of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c
ldc	leading dimension of matrix c

Returns

success

elpa_hermitian_multiply_a_h_a_real_double_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_a_h_a_real_double_impl	(		obj,
			uplo_a,
			uplo_c,
			ncb,
			a,
			b,
			ldb,
			ldbcols,
			c,
			ldc,
			ldccols )

elpa_hermitian_multiply_a_h_a_real_double_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a
obj%local_nrows	leading dimension of matrix a, set with class method objset("local_nrows",value)
b	matrix b
ldb	leading dimension of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c
ldc	leading dimension of matrix c

Returns

success

elpa_hermitian_multiply_a_h_a_real_single_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_a_h_a_real_single_impl ( obj, uplo_a, uplo_c, ncb, a, b, ldb, ldbcols, c, ldc, ldccols )

private

elpa_hermitian_multiply_real_a_h_a_single_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a
obj%local_nrows	leading dimension of matrix a, set with class method objset("local_nrows",value)
b	matrix b
ldb	leading dimension of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c
ldc	leading dimension of matrix c

Returns

success

elpa_hermitian_multiply_d_ptr_complex_double_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_d_ptr_complex_double_impl ( obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols )

private

elpa_hermitian_multiply_a_h_a_complex_double_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a, as a device pointer of type(c_ptr)
obj%local_ncols	leading dimension of matrix a, set with class method objset("local_nrows",value)
ldaCols	columns of matrix a
b	matrix b, as a device pointer of type(c_ptr)
ldb	leading dimension of matrix b
ldbCols	columns of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c, as a device_pointer of type(c_ptr)
ldc	leading dimension of matrix c

Returns

success

elpa_hermitian_multiply_d_ptr_complex_single_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_d_ptr_complex_single_impl ( obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols )

private

elpa_hermitian_multiply_d_ptr_complex_single_impl: Performs C : = A**H * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a, as a device pointer of type(c_ptr)
obj%local_ncols	leading dimension of matrix a, set with class method objset("local_nrows",value)
ldaCols	columns of matrix a
b	matrix b, as a device pointer of type(c_ptr)
ldb	leading dimension of matrix b
ldbCols	columns of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c, as a device pointer of type(c_ptr)
ldc	leading dimension of matrix c

Returns

success

elpa_hermitian_multiply_d_ptr_real_double_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_d_ptr_real_double_impl ( obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols )

private

elpa_hermitian_multiply_d_ptr_real_double_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a, as a device pointer of type(c_ptr)
obj%local_nrows	leading dimension of matrix a, set with class method objset("local_nrows",value)
b	matrix b, as a device pointer of type(c_ptr)
ldb	leading dimension of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c, as a device pointer of type(c_ptr)
ldc	leading dimension of matrix c

Returns

success

elpa_hermitian_multiply_d_ptr_real_single_impl()

function __mod_elpa_hermitian_multiply_F90__::elpa_hermitian_multiply_d_ptr_real_single_impl ( obj, uplo_a, uplo_c, ncb, adev, bdev, ldb, ldbcols, cdev, ldc, ldccols )

private

elpa_hermitian_multiply_real_d_ptr_single_impl: Performs C : = A**T * B where A is a square matrix (objna,objna) which is optionally upper or lower triangular B is a (objna,ncb) matrix C is a (objna,ncb) matrix where optionally only the upper or lower triangle may be computed

Parameters

uplo_a	'U' if A is upper triangular 'L' if A is lower triangular anything else if A is a full matrix Please note: This pertains to the original A (as set in the calling program) whereas the transpose of A is used for calculations If uplo_a is 'U' or 'L', the other triangle is not used at all, i.e. it may contain arbitrary numbers
uplo_c	'U' if only the upper diagonal part of C is needed 'L' if only the upper diagonal part of C is needed anything else if the full matrix C is needed Please note: Even when uplo_c is 'U' or 'L', the other triangle may be written to a certain extent, i.e. one shouldn't rely on the content there!
na	Number of rows/columns of A, number of rows of B and C
ncb	Number of columns of B and C
a	matrix a, as a device pointer of type(c_ptr)
obj%local_nrows	leading dimension of matrix a, set with class method objset("local_nrows",value)
b	matrix b, as a device pointer of type(c_ptr)
ldb	leading dimension of matrix b
nblk	blocksize of cyclic distribution, must be the same in both directions!
mpi_comm_rows	MPI communicator for rows
mpi_comm_cols	MPI communicator for columns
c	matrix c, as a device pointer of type(c_ptr)
ldc	leading dimension of matrix c

Returns

success