batchlas::backend::detail Namespace Reference

Functions

◆ syr2k_triangular_fast_path()

template<typename T>
bool batchlas::backend::detail::syr2k_triangular_fast_path ( const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & B,
const MatrixView< T, MatrixFormat::Dense > & C,
int n,
int k )

Definition at line 35 of file syr2k_triangular_tiles.hh.

◆ launch_syr2k_triangular_tiles()

template<typename T, bool TransOperand, bool AlignedFastPath>
Event batchlas::backend::detail::launch_syr2k_triangular_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & B,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
T beta,
Uplo uplo )

Definition at line 53 of file syr2k_triangular_tiles.hh.

◆ syr2k_triangular_tiles()

template<typename T>
Event batchlas::backend::detail::syr2k_triangular_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & B,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
T beta,
Uplo uplo,
Transpose transA )

Definition at line 278 of file syr2k_triangular_tiles.hh.

◆ real_part_of()

template<typename T>
T batchlas::backend::detail::real_part_of ( const T & value)
inline

Definition at line 35 of file syrk_gram_tiles.hh.

◆ launch_syrk_gram_tiles()

template<typename T, int NTile, int ThreadTile, int KC, bool TransOperand, bool Conjugate>
Event batchlas::backend::detail::launch_syrk_gram_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
T beta,
Uplo uplo )

Definition at line 47 of file syrk_gram_tiles.hh.

◆ syrk_gram_supported()

template<typename T>
bool batchlas::backend::detail::syrk_gram_supported ( const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & C,
Transpose transA,
bool conjugated )

Definition at line 248 of file syrk_gram_tiles.hh.

◆ syrk_gram_tiles()

template<typename T, bool Conjugate = false>
Event batchlas::backend::detail::syrk_gram_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
T beta,
Uplo uplo,
Transpose transA )

Definition at line 275 of file syrk_gram_tiles.hh.

◆ syrk_triangular_fast_path()

template<typename T>
bool batchlas::backend::detail::syrk_triangular_fast_path ( const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & C,
int n,
int k )

Definition at line 33 of file syrk_triangular_tiles.hh.

◆ launch_syrk_triangular_tiles()

template<typename T, bool TransOperand, bool AlignedFastPath>
Event batchlas::backend::detail::launch_syrk_triangular_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
T beta,
Uplo uplo )

Definition at line 49 of file syrk_triangular_tiles.hh.

◆ syrk_triangular_tiles()

template<typename T>
Event batchlas::backend::detail::syrk_triangular_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
T beta,
Uplo uplo,
Transpose transA )

Definition at line 271 of file syrk_triangular_tiles.hh.

◆ trmm_lanes()

int batchlas::backend::detail::trmm_lanes ( int tile)
inlineconstexpr

Definition at line 33 of file trmm_triangular_tiles.hh.

◆ launch_trmm_triangular_tiles()

template<typename T, int TileM>
Event batchlas::backend::detail::launch_trmm_triangular_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & B,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
Uplo uplo,
Transpose transA,
Diag diag )

Definition at line 42 of file trmm_triangular_tiles.hh.

◆ trmm_tiles_supported()

template<typename T>
bool batchlas::backend::detail::trmm_tiles_supported ( const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & B,
const MatrixView< T, MatrixFormat::Dense > & C,
Side side )

Definition at line 293 of file trmm_triangular_tiles.hh.

◆ trmm_row_tile()

template<typename T>
int batchlas::backend::detail::trmm_row_tile ( int m)
inline

Definition at line 319 of file trmm_triangular_tiles.hh.

◆ trmm_launch_tile_m()

template<typename T>
int batchlas::backend::detail::trmm_launch_tile_m ( int m)
inline

Definition at line 339 of file trmm_triangular_tiles.hh.

◆ trmm_tile_groups()

template<typename T>
std::int64_t batchlas::backend::detail::trmm_tile_groups ( std::int64_t m,
std::int64_t n )
inline

Definition at line 347 of file trmm_triangular_tiles.hh.

◆ trmm_triangular_tiles()

template<typename T>
Event batchlas::backend::detail::trmm_triangular_tiles ( Queue & ctx,
const MatrixView< T, MatrixFormat::Dense > & A,
const MatrixView< T, MatrixFormat::Dense > & B,
const MatrixView< T, MatrixFormat::Dense > & C,
T alpha,
Uplo uplo,
Transpose transA,
Diag diag )

Definition at line 353 of file trmm_triangular_tiles.hh.

Variables

◆ kGramMaxTile

int batchlas::backend::detail::kGramMaxTile = 128
inlineconstexpr

Definition at line 29 of file syrk_gram_tiles.hh.

◆ kTrmmTileN

int batchlas::backend::detail::kTrmmTileN = 128
inlineconstexpr

Definition at line 28 of file trmm_triangular_tiles.hh.

◆ kTrmmTileK

int batchlas::backend::detail::kTrmmTileK = 16
inlineconstexpr

Definition at line 29 of file trmm_triangular_tiles.hh.

Classes

class  Syr2kTriangularTilesKernel
class  SyrkGramTilesKernel
class  SyrkTriangularTilesKernel
class  TrmmTriangularTilesKernel