C++ API. More...
#include "experimental/group/fused_op/row_reduction_fused_op_xe.hpp"#include "experimental/group/reduction/row_reduce_store_xe.hpp"#include "experimental/kernel/reduction/api.hpp"#include "experimental/kernel/reduction/common.hpp"#include "experimental/kernel/reduction/config.hpp"

Go to the source code of this file.
Classes | |
| struct | gpu::xetla::kernel::xetla_row_reduction_t< dtype_in_, dtype_out_, dtype_acc_, reduction_attr_, gpu_arch::Xe, fused_op_t_ > |
| Is the row_reduction functor for Xe The idea is threads in group will cooperatively process matrix_m x wg_tile_n. More... | |
| struct | gpu::xetla::kernel::xetla_row_reduction_t< dtype_in_, dtype_out_, dtype_acc_, reduction_attr_, gpu_arch::Xe, fused_op_t_ >::arguments_t |
| struct | gpu::xetla::kernel::xetla_row_reduction_t< dtype_in_, dtype_out_, dtype_acc_, reduction_attr_, gpu_arch::Xe, fused_op_t_ >::get_barrier_count |
| struct | gpu::xetla::kernel::xetla_row_reduction_t< dtype_in_, dtype_out_, dtype_acc_, reduction_attr_, gpu_arch::Xe, fused_op_t_ >::get_slm_size |
Namespaces | |
| namespace | gpu |
| namespace | gpu::xetla |
| namespace | gpu::xetla::kernel |
C++ API.