Intel(R) Math Kernel Library for Deep Neural Networks (Intel(R) MKL-DNN)  0.21.0
Performance library for Deep Learning
Public Attributes | List of all members
mkldnn_pooling_desc_t Struct Reference

A descriptor of a pooling operation. More...

#include <mkldnn_types.h>

Collaboration diagram for mkldnn_pooling_desc_t:
Collaboration graph
[legend]

Public Attributes

mkldnn_primitive_kind_t primitive_kind
 The kind of primitive. More...
 
mkldnn_prop_kind_t prop_kind
 The kind of propagation. More...
 
mkldnn_alg_kind_t alg_kind
 The kind of pooling algorithm. More...
 
mkldnn_memory_desc_t src_desc
 Source memory descriptor. More...
 
mkldnn_memory_desc_t diff_src_desc
 Source gradient memory descriptor. More...
 
mkldnn_memory_desc_t dst_desc
 Destination memory descriptor. More...
 
mkldnn_memory_desc_t diff_dst_desc
 Destination gradient memory descriptor. More...
 
mkldnn_dims_t strides
 Pooling kernel strides for spatial dimensions. More...
 
mkldnn_dims_t kernel
 Pooling kernel spatial dimensions. More...
 
mkldnn_dims_t padding [2]
 Padding in each spatial dimension. More...
 
mkldnn_padding_kind_t padding_kind
 The kind of padding to use. More...
 
mkldnn_data_type_t accum_data_type
 The accumulator data type. More...
 

Detailed Description

A descriptor of a pooling operation.

Member Data Documentation

◆ primitive_kind

mkldnn_primitive_kind_t mkldnn_pooling_desc_t::primitive_kind

The kind of primitive.

Used for self-identifying the primitive descriptor. Must be mkldnn_pooling.

◆ prop_kind

mkldnn_prop_kind_t mkldnn_pooling_desc_t::prop_kind

The kind of propagation.

Possible values: mkldnn_forward_training, mkldnn_forward_inference, mkldnn_backward, and mkldnn_backward_data.

◆ alg_kind

mkldnn_alg_kind_t mkldnn_pooling_desc_t::alg_kind

The kind of pooling algorithm.

Possible values: mkldnn_pooling_max and mkldnn_pooling_avg.

◆ src_desc

mkldnn_memory_desc_t mkldnn_pooling_desc_t::src_desc

Source memory descriptor.

◆ diff_src_desc

mkldnn_memory_desc_t mkldnn_pooling_desc_t::diff_src_desc

Source gradient memory descriptor.

◆ dst_desc

mkldnn_memory_desc_t mkldnn_pooling_desc_t::dst_desc

Destination memory descriptor.

◆ diff_dst_desc

mkldnn_memory_desc_t mkldnn_pooling_desc_t::diff_dst_desc

Destination gradient memory descriptor.

◆ strides

mkldnn_dims_t mkldnn_pooling_desc_t::strides

Pooling kernel strides for spatial dimensions.

◆ kernel

mkldnn_dims_t mkldnn_pooling_desc_t::kernel

Pooling kernel spatial dimensions.

◆ padding

mkldnn_dims_t mkldnn_pooling_desc_t::padding[2]

Padding in each spatial dimension.

padding[0] is a padding in the beginning (padding_l), padding[1] is a padding in the end (padding_r).

◆ padding_kind

mkldnn_padding_kind_t mkldnn_pooling_desc_t::padding_kind

The kind of padding to use.

◆ accum_data_type

mkldnn_data_type_t mkldnn_pooling_desc_t::accum_data_type

The accumulator data type.

Initialized automatically.


The documentation for this struct was generated from the following file: