C++ API. More...
#include "experimental/kernel/data_transformer/api.hpp"#include "experimental/kernel/data_transformer/common.hpp"#include "experimental/kernel/data_transformer/config.hpp"#include "group/reduction/reduction_xe.hpp"

Go to the source code of this file.
Classes | |
| struct | gpu::xetla::kernel::xetla_data_transformer< dtype_in_, dtype_out_, dtype_compute_, data_transformer_attr_, mem_layout_in_, need_fp8_op, gpu_arch::Xe > |
| Is the data_transformer functor for Xe Each time, each thread will load sg_tile_m x sg_tile_n data into register and do the data convert. More... | |
| struct | gpu::xetla::kernel::xetla_data_transformer< dtype_in_, dtype_out_, dtype_compute_, data_transformer_attr_, mem_layout_in_, need_fp8_op, gpu_arch::Xe >::arguments_t |
| Arguments for gemm::run. More... | |
| struct | gpu::xetla::kernel::xetla_data_transformer< dtype_in_, dtype_out_, dtype_compute_, data_transformer_attr_, mem_layout_in_, need_fp8_op, gpu_arch::Xe >::get_barrier_count |
| struct | gpu::xetla::kernel::xetla_data_transformer< dtype_in_, dtype_out_, dtype_compute_, data_transformer_attr_, mem_layout_in_, need_fp8_op, gpu_arch::Xe >::get_slm_size |
Namespaces | |
| namespace | gpu |
| namespace | gpu::xetla |
| namespace | gpu::xetla::kernel |
C++ API.