multiplication : c <- beta c + alpha A b More...

Public Member Functions
subroutine	susmm (order, transA, nrhs, alpha, A, b, ldb, c, ldc, istat)
	Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of `transA`. More...

subroutine	dusmm (order, transA, nrhs, alpha, A, b, ldb, c, ldc, istat)
	Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of `transA`. More...

subroutine	cusmm (order, transA, nrhs, alpha, A, b, ldb, c, ldc, istat)
	Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of `transA`. More...

subroutine	zusmm (order, transA, nrhs, alpha, A, b, ldb, c, ldc, istat)
	Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of `transA`. More...

Detailed Description

multiplication : c <- beta c + alpha A b

Parameters

istat The return code will be written to istat (this is a Fortran routine): either 0 (success) or -1 (failure).

Member Function/Subroutine Documentation

◆ cusmm()

subroutine blas_sparse::usmm::cusmm	(	integer	order,
		integer	transA,
		integer	nrhs,
		complex(kind(1.e0))	alpha,
		integer	A,
		complex(kind(1.e0)), dimension (:,:)	b,
		integer	ldb,
		complex(kind(1.e0)), dimension (:,:)	c,
		integer	ldc,
		integer, intent(out)	istat
	)

Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of transA.

Parameters

order	layour of the dense array.
transA	Transposition operator for matrix A.
nrhs	Number of right hand side columns.
A	A valid matrix handle.
alpha	Value for $\alpha$ .
b	Dense vector b.
ldb	Leading dimension of b.
c	Dense vector c.
ldc	Leading dimension of c.

Note: By setting the blas_rsb_autotune_next_operation property via BLAS_ussp (at any time) the next multiplication routine call (either of BLAS_dusmv, BLAS_susmv, BLAS_zusmv, BLAS_cusmv, BLAS_dusmm, BLAS_susmm, BLAS_zusmm, BLAS_cusmm) will invoke autotuning before carrying out the effective operation. The tuning will take in account parameters like transposition, number of right hand sides, and scaling constants. By setting the blas_rsb_spmv_autotuning_on property via BLAS_ussp, the default number of executing threads for this matrix will be determined once, at matrix assembly time, and employed irrespective of the default threads count (different values for transposed and untransposed multiply). This can be overridden only by setting the RSB_NUM_THREADS environment variable.

See also: On the topic of autotuning, see also rsb_tune_spmm. If –enable-rsb-num-threads has been specified at configure time, the RSB_NUM_THREADS environment variable will override the number of executing threads specified by OMP_NUM_THREADS. (See also RSB_IO_WANT_EXECUTING_THREADS).

Returns: On success, 0 is returned; on error, -1.

Parameters

istat The return code will be written to istat (this is a Fortran routine): either 0 (success) or -1 (failure).

◆ dusmm()

subroutine blas_sparse::usmm::dusmm	(	integer	order,
		integer	transA,
		integer	nrhs,
		real(kind(1.d0))	alpha,
		integer	A,
		real(kind(1.d0)), dimension (:,:)	b,
		integer	ldb,
		real(kind(1.d0)), dimension (:,:)	c,
		integer	ldc,
		integer, intent(out)	istat
	)

Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of transA.

Parameters

order	layour of the dense array.
transA	Transposition operator for matrix A.
nrhs	Number of right hand side columns.
A	A valid matrix handle.
alpha	Value for $\alpha$ .
b	Dense vector b.
ldb	Leading dimension of b.
c	Dense vector c.
ldc	Leading dimension of c.

Note: By setting the blas_rsb_autotune_next_operation property via BLAS_ussp (at any time) the next multiplication routine call (either of BLAS_dusmv, BLAS_susmv, BLAS_zusmv, BLAS_cusmv, BLAS_dusmm, BLAS_susmm, BLAS_zusmm, BLAS_cusmm) will invoke autotuning before carrying out the effective operation. The tuning will take in account parameters like transposition, number of right hand sides, and scaling constants. By setting the blas_rsb_spmv_autotuning_on property via BLAS_ussp, the default number of executing threads for this matrix will be determined once, at matrix assembly time, and employed irrespective of the default threads count (different values for transposed and untransposed multiply). This can be overridden only by setting the RSB_NUM_THREADS environment variable.

See also: On the topic of autotuning, see also rsb_tune_spmm. If –enable-rsb-num-threads has been specified at configure time, the RSB_NUM_THREADS environment variable will override the number of executing threads specified by OMP_NUM_THREADS. (See also RSB_IO_WANT_EXECUTING_THREADS).

Returns: On success, 0 is returned; on error, -1.

Parameters

istat The return code will be written to istat (this is a Fortran routine): either 0 (success) or -1 (failure).

◆ susmm()

subroutine blas_sparse::usmm::susmm	(	integer	order,
		integer	transA,
		integer	nrhs,
		real(kind(1.e0))	alpha,
		integer	A,
		real(kind(1.e0)), dimension (:,:)	b,
		integer	ldb,
		real(kind(1.e0)), dimension (:,:)	c,
		integer	ldc,
		integer, intent(out)	istat
	)

Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of transA.

Parameters

order	layour of the dense array.
transA	Transposition operator for matrix A.
nrhs	Number of right hand side columns.
A	A valid matrix handle.
alpha	Value for $\alpha$ .
b	Dense vector b.
ldb	Leading dimension of b.
c	Dense vector c.
ldc	Leading dimension of c.

Note: By setting the blas_rsb_autotune_next_operation property via BLAS_ussp (at any time) the next multiplication routine call (either of BLAS_dusmv, BLAS_susmv, BLAS_zusmv, BLAS_cusmv, BLAS_dusmm, BLAS_susmm, BLAS_zusmm, BLAS_cusmm) will invoke autotuning before carrying out the effective operation. The tuning will take in account parameters like transposition, number of right hand sides, and scaling constants. By setting the blas_rsb_spmv_autotuning_on property via BLAS_ussp, the default number of executing threads for this matrix will be determined once, at matrix assembly time, and employed irrespective of the default threads count (different values for transposed and untransposed multiply). This can be overridden only by setting the RSB_NUM_THREADS environment variable.

See also: On the topic of autotuning, see also rsb_tune_spmm. If –enable-rsb-num-threads has been specified at configure time, the RSB_NUM_THREADS environment variable will override the number of executing threads specified by OMP_NUM_THREADS. (See also RSB_IO_WANT_EXECUTING_THREADS).

Returns: On success, 0 is returned; on error, -1.

Parameters

istat The return code will be written to istat (this is a Fortran routine): either 0 (success) or -1 (failure).

◆ zusmm()

subroutine blas_sparse::usmm::zusmm	(	integer	order,
		integer	transA,
		integer	nrhs,
		complex(kind(1.d0))	alpha,
		integer	A,
		complex(kind(1.d0)), dimension (:,:)	b,
		integer	ldb,
		complex(kind(1.d0)), dimension (:,:)	c,
		integer	ldc,
		integer, intent(out)	istat
	)

Multiply by a dense matrix (aka multi-vector). Either of $C \leftarrow \alpha AB+C,$ $C \leftarrow \alpha A^T B+C,$ $C \leftarrow \alpha A^H B+C$ , depending on the value of transA.

Parameters

order	layour of the dense array.
transA	Transposition operator for matrix A.
nrhs	Number of right hand side columns.
A	A valid matrix handle.
alpha	Value for $\alpha$ .
b	Dense vector b.
ldb	Leading dimension of b.
c	Dense vector c.
ldc	Leading dimension of c.

Note: By setting the blas_rsb_autotune_next_operation property via BLAS_ussp (at any time) the next multiplication routine call (either of BLAS_dusmv, BLAS_susmv, BLAS_zusmv, BLAS_cusmv, BLAS_dusmm, BLAS_susmm, BLAS_zusmm, BLAS_cusmm) will invoke autotuning before carrying out the effective operation. The tuning will take in account parameters like transposition, number of right hand sides, and scaling constants. By setting the blas_rsb_spmv_autotuning_on property via BLAS_ussp, the default number of executing threads for this matrix will be determined once, at matrix assembly time, and employed irrespective of the default threads count (different values for transposed and untransposed multiply). This can be overridden only by setting the RSB_NUM_THREADS environment variable.

See also: On the topic of autotuning, see also rsb_tune_spmm. If –enable-rsb-num-threads has been specified at configure time, the RSB_NUM_THREADS environment variable will override the number of executing threads specified by OMP_NUM_THREADS. (See also RSB_IO_WANT_EXECUTING_THREADS).

Returns: On success, 0 is returned; on error, -1.

Parameters

istat The return code will be written to istat (this is a Fortran routine): either 0 (success) or -1 (failure).

The documentation for this interface was generated from the following file:

rsb_blas_sparse.F90

Public Member Functions

Detailed Description

Member Function/Subroutine Documentation

◆ cusmm()

◆ dusmm()

◆ susmm()

◆ zusmm()