![]() |
cuLite v0.3.1
A lite CUDA C++ Interface
|
Import and aliasing of CLA3P types into the cuLite namespace. More...
Typedefs | |
| template<typename T_obj> | |
| using | culite::Guard = ::cla3p::Guard<T_obj> |
| Immutable object wrapper. | |
| template<typename T_Int> | |
| using | culite::MatrixMeta = ::cla3p::MatrixMeta<T_Int> |
| Matrix metadata container. | |
| template<typename T_Int> | |
| using | culite::Meta1D = ::cla3p::Meta1D<T_Int> |
| One-dimensional metadata container. | |
| template<typename T_Int> | |
| using | culite::Meta2D = ::cla3p::Meta2D<T_Int> |
| Two-dimensional metadata container. | |
| using | culite::prop_t = ::cla3p::prop_t |
| Matrix algebraic property enumeration. | |
| using | culite::uplo_t = ::cla3p::uplo_t |
| Matrix fill type enumeration. | |
| using | culite::op_t = ::cla3p::op_t |
| Virtual matrix operation enumeration. | |
| using | culite::diag_t = ::cla3p::diag_t |
| Diagonal element type enumeration. | |
| using | culite::side_t = ::cla3p::side_t |
| Matrix operand positioning enumeration. | |
| using | culite::decomp_t = ::cla3p::decomp_t |
| Matrix decomposition method enumeration. | |
| using | culite::svdPolicy_t = ::cla3p::svdPolicy_t |
| Singular vector computation policy enumeration. | |
| using | culite::eigRange_t = ::cla3p::eigRange_t |
| Eigenvalue computation range enumeration. | |
| using | culite::Property = ::cla3p::Property |
| The property class. | |
Import and aliasing of CLA3P types into the cuLite namespace.
This module documents the types and enumerations that cuLite imports from the CLA3P library to maintain API consistency between CPU and GPU implementations. Includes matrix properties, operation flags, decomposition policies, and generic container types.
| using culite::Guard = ::cla3p::Guard<T_obj> |
Immutable object wrapper.
| T_obj | The object type to guard. |
Alias of cla3p::Guard into the culite namespace. Provides a read-only view of an object, preventing modifications to its contents.
| using culite::MatrixMeta = ::cla3p::MatrixMeta<T_Int> |
Matrix metadata container.
| T_Int | Integer type for dimension values. |
Alias of cla3p::MatrixMeta into the culite namespace. Extends culite::Meta2D with matrix property information, storing dimensions and structural properties (e.g., symmetric, triangular).
| using culite::Meta1D = ::cla3p::Meta1D<T_Int> |
One-dimensional metadata container.
| T_Int | Integer type for storing the dimension value. |
Alias of cla3p::Meta1D into the culite namespace. Provides storage and accessors for vector size.
| using culite::Meta2D = ::cla3p::Meta2D<T_Int> |
Two-dimensional metadata container.
| T_Int | Integer type for storing dimension values. |
Alias of cla3p::Meta2D into the culite namespace. Provides storage and accessors for matrix dimensions (rows and columns).
| using culite::prop_t = ::cla3p::prop_t |
Matrix algebraic property enumeration.
Alias of cla3p::prop_t into the culite namespace. Defines the algebraic properties of matrices used to optimize computational operations. Each property type imposes specific structural constraints and enables specialized algorithms.
| using culite::uplo_t = ::cla3p::uplo_t |
Matrix fill type enumeration.
Alias of cla3p::uplo_t into the culite namespace. Specifies which triangular portion of a matrix is stored and utilized. This enumeration is used to define storage efficiency for symmetric, Hermitian, and triangular matrices where only one triangular portion needs to be stored.
| using culite::op_t = ::cla3p::op_t |
Virtual matrix operation enumeration.
Alias of cla3p::op_t into the culite namespace. Specifies virtual operations to be applied to matrices in computational routines. These operations are not explicitly performed; instead, the operation flag is passed to optimized algorithms that implicitly handle the transformation.
| using culite::diag_t = ::cla3p::diag_t |
Diagonal element type enumeration.
Alias of cla3p::diag_t into the culite namespace. Specifies whether the diagonal of a triangular matrix consists of unit elements or arbitrary values. This affects storage requirements and computational algorithms.
| using culite::side_t = ::cla3p::side_t |
Matrix operand positioning enumeration.
Alias of cla3p::side_t into the culite namespace. Specifies the position (left or right) of a particular matrix operand in binary matrix operations such as triangular solves and matrix multiplication.
| using culite::decomp_t = ::cla3p::decomp_t |
Matrix decomposition method enumeration.
Alias of cla3p::decomp_t into the culite namespace. Specifies the factorization algorithm to be used for matrix decomposition. The choice of decomposition method depends on matrix properties and computational requirements.
Singular vector computation policy enumeration.
Alias of cla3p::svdPolicy_t into the culite namespace. Specifies the extent of singular vector computation in Singular Value Decomposition (SVD). Different policies trade off computational cost against the completeness of the decomposition.
Eigenvalue computation range enumeration.
Alias of cla3p::eigRange_t into the culite namespace. Specifies which subset of eigenvalues to compute in eigenvalue decomposition routines. This allows selective computation to improve performance when only specific eigenvalues are needed, rather than computing the full spectrum.
| using culite::Property = ::cla3p::Property |
The property class.
Alias of cla3p::Property into the culite namespace. Used to define matrix properties. Constructing a matrix with property can accelerate certain operations, but at the same time imposes equivalent constraints on others.