tile_op_functor.hpp File Reference

C++ API. More...

#include "subgroup/tile/api.hpp"
#include "subgroup/tile/common.hpp"
#include "subgroup/tile/impl/load_xe.hpp"
#include "subgroup/tile/impl/payload_xe.hpp"
#include "subgroup/tile/impl/prefetch_xe.hpp"
#include "subgroup/tile/impl/store_xe.hpp"

Include dependency graph for tile_op_functor.hpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	gpu::xetla::subgroup::none_op_t
	Is none op functor, for placeholder purpose. More...
struct	gpu::xetla::subgroup::none_op_t::arguments_t
struct	gpu::xetla::subgroup::relu_op_t
	Is the element-wise relu op functor. More...
struct	gpu::xetla::subgroup::relu_op_t::arguments_t
struct	gpu::xetla::subgroup::tanh_op_t
	Is the element-wise tanh op functor. More...
struct	gpu::xetla::subgroup::tanh_op_t::arguments_t
struct	gpu::xetla::subgroup::sigmoid_op_t
	Is the element-wise sigmoid op functor. More...
struct	gpu::xetla::subgroup::sigmoid_op_t::arguments_t
struct	gpu::xetla::subgroup::gelu_fwd_op_t
	Is the element-wise gelu inference forward op functor. More...
struct	gpu::xetla::subgroup::gelu_fwd_op_t::arguments_t
struct	gpu::xetla::subgroup::gelu_fwd_w_op_t< dtype_out, arch_tag, enable >
	Is the element-wise gelu training forward op functor. More...
struct	gpu::xetla::subgroup::gelu_fwd_w_op_t< dtype_out_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the element-wise gelu training forward op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::gelu_fwd_w_op_t< dtype_out_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::gelu_bwd_op_t< dtype_in, arch_tag, enable >
	Is the element-wise gelu backward op functor. More...
struct	gpu::xetla::subgroup::gelu_bwd_op_t< dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the element-wise gelu backward op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::gelu_bwd_op_t< dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::bias_add_op_t< dtype_bias, arch_tag, enable >
	Is the bias_add op functor. More...
struct	gpu::xetla::subgroup::bias_add_op_t< dtype_bias_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the bias_add op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::bias_add_op_t< dtype_bias_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::scale_v_offset_v_op_t< scale_dtype, offset_dtype, arch_tag, enable >
	Is MatAcc * vector scale + vector offset. More...
struct	gpu::xetla::subgroup::scale_v_offset_v_op_t< scale_dtype_, offset_dtype_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the scale_v_offset_v op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::scale_v_offset_v_op_t< scale_dtype_, offset_dtype_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::scale_v_op_t< scale_dtype, arch_tag, enable >
	Is MatAcc * vector scale. More...
struct	gpu::xetla::subgroup::scale_v_op_t< scale_dtype_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the scale_v op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::scale_v_op_t< scale_dtype_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::elemwise_reduce_op_t< reduce_kind, dtype_in, arch_tag, enable >
	Is the element-wise reduce op functor. More...
struct	gpu::xetla::subgroup::elemwise_reduce_op_t< reduce_kind_, dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the element-wise reduce op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::elemwise_reduce_op_t< reduce_kind_, dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::elemwise_reduce_op_stream_k_t< reduce_kind, dtype_in, arch_tag >
	Is the element-wise reduce op functor, specialized for stream_k dispatch Load partial sum from scratchspace Reduce in GRF Store zero to scratchspace Do these steps with smaller tiles to minimize GRF pressure. More...
struct	gpu::xetla::subgroup::elemwise_reduce_op_stream_k_t< reduce_kind_, dtype_in_, gpu_arch::Xe >
	Is the element-wise reduce op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::elemwise_reduce_op_stream_k_t< reduce_kind_, dtype_in_, gpu_arch::Xe >::arguments_t
struct	gpu::xetla::subgroup::dropout_op_t< dtype_mask, arch_tag, enable >
	Is the dropout op functor. More...
struct	gpu::xetla::subgroup::dropout_op_t< dtype_mask_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the dropout op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::dropout_op_t< dtype_mask_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::rng_dropout_op_t< dtype_mask, arch_tag, enable >
	Is the random number generator and dropout op functor. More...
struct	gpu::xetla::subgroup::rng_dropout_op_t< dtype_mask_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the random number generator and dropout op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::rng_dropout_op_t< dtype_mask_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::scalar_mul_op_t< dtype_in, arch_tag, enable >
	Is the scalar_multiply op functor. More...
struct	gpu::xetla::subgroup::scalar_mul_op_t< dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the scalar_multiply op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::scalar_mul_op_t< dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t
struct	gpu::xetla::subgroup::linear_op_t< dtype_in, arch_tag, enable >
	Is the linear_op functor. More...
struct	gpu::xetla::subgroup::linear_op_t< dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >
	Is the linear_op functor, specialized for Xe architecture. More...
struct	gpu::xetla::subgroup::linear_op_t< dtype_in_, arch_tag, std::enable_if_t<(arch_tag==gpu_arch::Xe)> >::arguments_t

Namespaces
namespace	gpu
namespace	gpu::xetla
namespace	gpu::xetla::subgroup

Detailed Description

C++ API.