hyhound.tpp File Reference

#include <batmat/assume.hpp>
#include <batmat/linalg/micro-kernels/hyhound.hpp>
#include <batmat/linalg/uview.hpp>
#include <batmat/loop.hpp>
#include <batmat/ops/cneg.hpp>
#include <batmat/ops/rotate.hpp>
#include <guanaqo/trace.hpp>
#include <type_traits>
#include <utility>

Go to the source code of this file.

Namespaces
namespace	batmat
namespace	batmat::linalg
namespace	batmat::linalg::micro_kernels
namespace	batmat::linalg::micro_kernels::hyhound
namespace	batmat::linalg::micro_kernels::hyhound::detail

Macros
#define	UNROLL_FOR(...)

Functions
template<class T, class Abi, KernelConfig Conf, index_t R, StorageOrder OL, StorageOrder OA>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_diag_microkernel (index_t kA, triangular_accessor< T, Abi, SizeR< T, Abi > > W, uview< T, Abi, OL > L, uview< T, Abi, OA > A, uview< const T, Abi, StorageOrder::ColMajor > diag) noexcept
template<class T, class Abi, KernelConfig Conf, index_t R, StorageOrder OL, StorageOrder OA>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_full_microkernel (index_t kA, uview< T, Abi, OL > L, uview< T, Abi, OA > A, uview< const T, Abi, StorageOrder::ColMajor > diag) noexcept
template<class T, class Abi, int S>
auto	batmat::linalg::micro_kernels::hyhound::detail::rotate (datapar::simd< T, Abi > x, std::integral_constant< int, S >)
template<class T, class Abi>
auto	batmat::linalg::micro_kernels::hyhound::detail::rotate (datapar::simd< T, Abi > x, int s)
template<class T, class Abi, KernelConfig Conf, index_t R, index_t S, StorageOrder OL, StorageOrder OA, StorageOrder OB>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_tail_microkernel (index_t kA_in_offset, index_t kA_in, index_t k, triangular_accessor< const T, Abi, SizeR< T, Abi > > W, uview< T, Abi, OL > L, uview< const T, Abi, OA > A_in, uview< T, Abi, OA > A_out, uview< const T, Abi, OB > B, uview< const T, Abi, StorageOrder::ColMajor > diag, Structure struc_L, int rotate_A) noexcept
template<class T, class Abi, KernelConfig Conf = {}, StorageOrder OL = StorageOrder::ColMajor, StorageOrder OA = StorageOrder::ColMajor>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_register (const view< T, Abi, OL > L, const view< T, Abi, OA > A, const view< const T, Abi > D) noexcept
	Block hyperbolic Householder factorization update using register blocking.
template<class T, class Abi, KernelConfig Conf = {}, StorageOrder OL = StorageOrder::ColMajor, StorageOrder OA = StorageOrder::ColMajor>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_register (const view< T, Abi, OL > L, const view< T, Abi, OA > A, const view< const T, Abi > D, const view< T, Abi > W) noexcept
	Block hyperbolic Householder factorization update using register blocking.
template<class T, class Abi, KernelConfig Conf = {}, StorageOrder OL = StorageOrder::ColMajor, StorageOrder OA = StorageOrder::ColMajor>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_apply_register (const view< T, Abi, OL > L, const view< const T, Abi, OA > Ain, const view< T, Abi, OA > Aout, const view< const T, Abi, OA > B, const view< const T, Abi > D, const view< const T, Abi > W, index_t kA_in_offset) noexcept
	Apply a block hyperbolic Householder transformation.
template<class T, class Abi, StorageOrder OL1 = StorageOrder::ColMajor, StorageOrder OA1 = StorageOrder::ColMajor, StorageOrder OL2 = StorageOrder::ColMajor, StorageOrder OA2 = StorageOrder::ColMajor, KernelConfig Conf = {}>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_2_register (const view< T, Abi, OL1 > L11, const view< T, Abi, OA1 > A1, const view< T, Abi, OL2 > L21, const view< T, Abi, OA2 > A2, const view< const T, Abi > D) noexcept
	Same as hyhound_diag_register but for two block rows at once.
template<class T, class Abi, StorageOrder OL = StorageOrder::ColMajor, StorageOrder OW = StorageOrder::ColMajor, StorageOrder OY = StorageOrder::ColMajor, StorageOrder OU = StorageOrder::ColMajor, KernelConfig Conf = {}>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_cyclic_register (const view< T, Abi, OL > L11, const view< T, Abi, OW > A1, const view< T, Abi, OY > L21, const view< const T, Abi, OW > A22, const view< T, Abi, OW > A2_out, const view< T, Abi, OU > L31, const view< const T, Abi, OW > A31, const view< T, Abi, OW > A3_out, const view< const T, Abi > D) noexcept
	Performs a factorization update of the following matrix:
template<class T, class Abi, StorageOrder OL = StorageOrder::ColMajor, StorageOrder OA = StorageOrder::ColMajor, StorageOrder OLu = StorageOrder::ColMajor, StorageOrder OAu = StorageOrder::ColMajor, KernelConfig Conf = {}>
void	batmat::linalg::micro_kernels::hyhound::hyhound_diag_riccati_register (const view< T, Abi, OL > L11, const view< T, Abi, OA > A1, const view< T, Abi, OL > L21, const view< const T, Abi, OA > A2, const view< T, Abi, OA > A2_out, const view< T, Abi, OLu > Lu1, const view< T, Abi, OAu > Au_out, const view< const T, Abi > D, bool shift_A_out) noexcept
	Performs a factorization update of the following matrix:

Macro Definition Documentation

UNROLL_FOR

#define UNROLL_FOR

(

...

)

Value:

BATMAT_FULLY_UNROLLED_FOR (__VA_ARGS__)

Definition at line 13 of file hyhound.tpp.