tile_size x tile_size register-tile micro-kernels of the TiledLUpp solvers: the shared kernels plus the LU column elimination. More...

#include <tile_operations.hpp>

Inheritance diagram for tdls::TiledLUppTileOperations< T, tile_size, unroll_inner >:

[legend]

Collaboration diagram for tdls::TiledLUppTileOperations< T, tile_size, unroll_inner >:

[legend]

Static Public Member Functions
template<int row_extent, int col_extent>
TDLS_HOST_DEVICE static TDLS_FORCEINLINE constexpr void	eliminate_column (T *TDLS_RESTRICT t, int k) noexcept
	Gaussian elimination of column k inside the diagonal tile.

Static Public Member Functions inherited from tdls::TileOperations< T, tile_size, unroll_inner >
template<int k_extent>
TDLS_HOST_DEVICE static TDLS_FORCEINLINE constexpr void	swap_rows (T *TDLS_RESTRICT t, int k, int r) noexcept
	Row swap k <-> r inside the KExKE active part of the tile, compile-time indexed on both sides.

template<int diag_extent, int col_extent>
TDLS_HOST_DEVICE static TDLS_FORCEINLINE constexpr void	trsm_left_unit (const T TDLS_RESTRICT lu, T TDLS_RESTRICT B) noexcept
	B := L^-1 B, with L the unit lower part of the factored diagonal tile. L is KDxKD, B is KDxC.

template<int diag_extent, int row_extent>
TDLS_HOST_DEVICE static TDLS_FORCEINLINE constexpr void	trsm_right (const T TDLS_RESTRICT lu, T TDLS_RESTRICT B) noexcept
	B := B U^-1, with U the upper part of the factored diagonal tile. U is KDxKD, B is RxKD.

template<int row_extent, int col_extent, int k_extent>
TDLS_HOST_DEVICE static TDLS_FORCEINLINE constexpr void	gemm_sub (T TDLS_RESTRICT Ct, const T TDLS_RESTRICT At, const T *TDLS_RESTRICT Bt) noexcept
	Ct -= At*Bt with per-element dot-product accumulation. At is RxK, Bt is KxC, Ct is RxC.

Detailed Description

template<typename T, int tile_size, bool unroll_inner>
struct tdls::TiledLUppTileOperations< T, tile_size, unroll_inner >

tile_size x tile_size register-tile micro-kernels of the TiledLUpp solvers: the shared kernels plus the LU column elimination.

Template Parameters

T	scalar type
tile_size	tile size (int, row stride of the register tiles)
unroll_inner	unroll knob, forwarded from the TiledLUpp solver configuration

Member Function Documentation

◆ eliminate_column()

template<typename T , int tile_size, bool unroll_inner>

template<int row_extent, int col_extent>

TDLS_HOST_DEVICE static TDLS_FORCEINLINE constexpr void tdls::TiledLUppTileOperations< T, tile_size, unroll_inner >::eliminate_column	(	T *TDLS_RESTRICT	t,
		int	k )

inlinestaticconstexprnoexcept

Gaussian elimination of column k inside the diagonal tile.

Scales the sub-column by 1/pivot and updates the trailing block. Extents <row_extent, col_extent> bound the active part of the tile. The RECIPROCAL of the pivot is stored at the diagonal slot: every downstream consumer (trsm_right, backward substitution, out-of-tile replays) multiplies instead of dividing. The division is paid once here, where it already had to happen.

Template Parameters

row_extent	active row extent of the tile
col_extent	active column extent of the tile

Parameters

[in,out]	t	register tile
[in]	k	column to eliminate

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

include/tdls/solvers/tiled_lupp/tile_operations.hpp

Static Public Member Functions

Detailed Description

Member Function Documentation

◆ eliminate_column()