gko::matrix::Sellp#

Sliced ELL with permutation. Partitions rows into fixed-size slices, pads each slice independently to its own longest row, and optionally permutes rows within a slice to balance work. A middle ground between ELL’s regular layout and CSR’s storage efficiency.

template<typename ValueType = default_precision, typename IndexType = int32>
class Sellp #

Inherits from

  • public gko::EnableLinOp<Sellp<default_precision, int32>>

  • public ConvertibleTo<Sellp<next_precision<default_precision>, int32>>

  • public ConvertibleTo<Sellp<next_precision<default_precision, 2>, int32>>

  • public ConvertibleTo<Sellp<next_precision<default_precision, 3>, int32>>

  • public ConvertibleTo<Dense<default_precision>>

  • public ConvertibleTo<Csr<default_precision, int32>>

  • public gko::DiagonalExtractable<default_precision>

  • public gko::ReadableFromMatrixData<default_precision, int32>

  • public gko::WritableToMatrixData<default_precision, int32>

  • public gko::EnableAbsoluteComputation<remove_complex<Sellp<default_precision, int32>>>

SELL-P is a matrix format similar to ELL format. The difference is that SELL-P format divides rows into smaller slices and store each slice with ELL format.

This implementation uses the column index value invalid_index<IndexType>() to mark padding entries that are not part of the sparsity pattern.

Template Parameters:

  • ValueType – precision of matrix elements

  • IndexType – precision of matrix indexes

Public Functions

virtual std::unique_ptr<Diagonal<ValueType>> extract_diagonal(
) const override#

Extracts the diagonal entries of the matrix into a vector.

Parameters:

diag – the vector into which the diagonal will be written

virtual std::unique_ptr<absolute_type> compute_absolute(
) const override#

Gets the AbsoluteLinOp

Returns:

a pointer to the new absolute object

virtual void compute_absolute_inplace() override#

Compute absolute inplace on each element.

inline value_type *get_values() noexcept#

Returns the values of the matrix.

Returns:

the values of the matrix.

inline const value_type *get_const_values() const noexcept#

Returns the values of the matrix.

Note

This is the constant version of the function, which can be significantly more memory efficient than the non-constant version, so always prefer this version.

Returns:

the values of the matrix.

inline index_type *get_col_idxs() noexcept#

Returns the column indexes of the matrix.

Returns:

the column indexes of the matrix.

inline const index_type *get_const_col_idxs() const noexcept#

Returns the column indexes of the matrix.

Note

This is the constant version of the function, which can be significantly more memory efficient than the non-constant version, so always prefer this version.

Returns:

the column indexes of the matrix.

inline size_type *get_slice_lengths() noexcept#

Returns the lengths(columns) of slices.

Returns:

the lengths(columns) of slices.

inline const size_type *get_const_slice_lengths() const noexcept#

Returns the lengths(columns) of slices.

Note

This is the constant version of the function, which can be significantly more memory efficient than the non-constant version, so always prefer this version.

Returns:

the lengths(columns) of slices.

inline size_type *get_slice_sets() noexcept#

Returns the offsets of slices.

Returns:

the offsets of slices.

inline const size_type *get_const_slice_sets() const noexcept#

Returns the offsets of slices.

Note

This is the constant version of the function, which can be significantly more memory efficient than the non-constant version, so always prefer this version.

Returns:

the offsets of slices.

inline size_type get_slice_size() const noexcept#

Returns the size of a slice.

Returns:

the size of a slice.

inline size_type get_stride_factor() const noexcept#

Returns the stride factor(t) of SELL-P.

Returns:

the stride factor(t) of SELL-P.

inline size_type get_total_cols() const noexcept#

Returns the total column number.

Returns:

the total column number.

inline size_type get_num_stored_elements() const noexcept#

Returns the number of elements explicitly stored in the matrix.

Returns:

the number of elements explicitly stored in the matrix

inline value_type &val_at(
size_type row,
size_type slice_set,
size_type idx,
) noexcept#

Returns the idx-th non-zero element of the row-th row with slice_set slice set.

Note

the method has to be called on the same Executor the matrix is stored at (e.g. trying to call this method on a GPU matrix from the CPU results in a runtime error)

Parameters:

  • row – the row of the requested element in the slice

  • slice_set – the slice set of the slice

  • idx – the idx-th stored element of the row in the slice

inline value_type val_at(
size_type row,
size_type slice_set,
size_type idx,
) const noexcept#

Returns the idx-th non-zero element of the row-th row with slice_set slice set.

Note

the method has to be called on the same Executor the matrix is stored at (e.g. trying to call this method on a GPU matrix from the CPU results in a runtime error)

Parameters:

  • row – the row of the requested element in the slice

  • slice_set – the slice set of the slice

  • idx – the idx-th stored element of the row in the slice

inline index_type &col_at(
size_type row,
size_type slice_set,
size_type idx,
) noexcept#

Returns the idx-th column index of the row-th row with slice_set slice set.

Note

the method has to be called on the same Executor the matrix is stored at (e.g. trying to call this method on a GPU matrix from the CPU results in a runtime error)

Parameters:

  • row – the row of the requested element in the slice

  • slice_set – the slice set of the slice

  • idx – the idx-th stored element of the row in the slice

inline index_type col_at(
size_type row,
size_type slice_set,
size_type idx,
) const noexcept#

Returns the idx-th column index of the row-th row with slice_set slice set.

Note

the method has to be called on the same Executor the matrix is stored at (e.g. trying to call this method on a GPU matrix from the CPU results in a runtime error)

Parameters:

  • row – the row of the requested element in the slice

  • slice_set – the slice set of the slice

  • idx – the idx-th stored element of the row in the slice

device_view get_device_view()#

get the non-owning device view

const_device_view get_const_device_view() const#

get the const non-owning device view

Sellp &operator=(const Sellp&)#

Copy-assigns a Sellp matrix. Preserves the executor, copies the data and parameters.

Sellp &operator=(Sellp&&)#

Move-assigns a Sellp matrix. Preserves the executor, moves the data and parameters. The moved-from object is empty (0x0 with valid slice_sets and unchanged parameters).

Sellp(const Sellp&)#

Copy-assigns a Sellp matrix. Inherits the executor, copies the data and parameters.

Sellp(Sellp&&)#

Move-assigns a Sellp matrix. Inherits the executor, moves the data and parameters. The moved-from object is empty (0x0 with valid slice_sets and unchanged parameters).

Public Static Functions

static std::unique_ptr<Sellp> create(
std::shared_ptr<const Executor> exec,
const dim<2> &size = {},
size_type total_cols = 0,
)#

Creates an uninitialized Sellp matrix of the specified size. (The slice_size and stride_factor are set to the default values.)

Parameters:

  • exec – Executor associated to the matrix

  • size – size of the matrix

  • total_cols – number of the sum of all cols in every slice.

Returns:

A smart pointer to the newly created matrix.

static std::unique_ptr<Sellp> create(
std::shared_ptr<const Executor> exec,
const dim<2> &size,
size_type slice_size,
size_type stride_factor,
size_type total_cols,
)#

Creates an uninitialized Sellp matrix of the specified size.

Parameters:

  • exec – Executor associated to the matrix

  • size – size of the matrix

  • slice_size – number of rows in each slice

  • stride_factor – factor for the stride in each slice (strides should be multiples of the stride_factor)

  • total_cols – number of the sum of all cols in every slice.

Returns:

A smart pointer to the newly created matrix.