batmat main
Batched linear algebra routines
Loading...
Searching...
No Matches
gemm.hpp File Reference
#include <batmat/linalg/structure.hpp>
#include <batmat/linalg/uview.hpp>
#include <batmat/lut.hpp>
#include <batmat/micro-kernels/gemm/export.h>
#include <batmat/platform/platform.hpp>
#include <optional>

Go to the source code of this file.

Classes

struct  batmat::linalg::micro_kernels::gemm::KernelConfig

Namespaces

namespace  batmat
namespace  batmat::linalg
namespace  batmat::linalg::micro_kernels
namespace  batmat::linalg::micro_kernels::gemm
namespace  batmat::linalg::micro_kernels::gemm::detail

Functions

template<class T, class Abi, KernelConfig Conf, index_t RowsReg, index_t ColsReg, StorageOrder OA, StorageOrder OB, StorageOrder OC, StorageOrder OD>
void batmat::linalg::micro_kernels::gemm::gemm_copy_microkernel (const uview< const T, Abi, OA > A, const uview< const T, Abi, OB > B, const std::optional< uview< const T, Abi, OC > > C, const uview< T, Abi, OD > D, const index_t k) noexcept
 Generalized matrix multiplication D = C ± A⁽ᵀ⁾ B⁽ᵀ⁾. Single register block.
template<class T, class Abi, KernelConfig Conf, StorageOrder OA, StorageOrder OB, StorageOrder OC, StorageOrder OD>
void batmat::linalg::micro_kernels::gemm::gemm_copy_register (const view< const T, Abi, OA > A, const view< const T, Abi, OB > B, const std::optional< view< const T, Abi, OC > > C, const view< T, Abi, OD > D) noexcept
 Generalized matrix multiplication D = C ± A⁽ᵀ⁾ B⁽ᵀ⁾. Using register blocking.

Variables

template<class T, class Abi>
constexpr index_t batmat::linalg::micro_kernels::gemm::ColsReg = RowsReg<T, Abi>
template<class T, class Abi, KernelConfig Conf, StorageOrder OA, StorageOrder OB, StorageOrder OC, StorageOrder OD>
constexpr auto batmat::linalg::micro_kernels::gemm::detail::gemm_copy_lut
template<class T, class Abi, KernelConfig Conf, StorageOrder OA, StorageOrder OB, StorageOrder OC, StorageOrder OD>
const constinit decltype(detail::gemm_copy_lut< T, Abi, Conf, OA, OB, OC, OD >) batmat::linalg::micro_kernels::gemm::gemm_copy_lut = detail::gemm_copy_lut<T, Abi, Conf, OA, OB, OC, OD>

Class Documentation

◆ batmat::linalg::micro_kernels::gemm::KernelConfig

struct batmat::linalg::micro_kernels::gemm::KernelConfig
Class Members
bool negate = false
int shift_A = 0
int rotate_B = 0
int rotate_C = 0
int rotate_D = rotate_C
int mask_D = rotate_D
MatrixStructure struc_A = MatrixStructure::General
MatrixStructure struc_B = MatrixStructure::General
MatrixStructure struc_C = MatrixStructure::General