Deep Neural Network Library (DNNL)  1.1.3
Performance library for Deep Learning
Public Attributes | List of all members
dnnl_convolution_desc_t Struct Reference

A descriptor of a convolution operation. More...

#include <dnnl_types.h>

Collaboration diagram for dnnl_convolution_desc_t:
Collaboration graph
[legend]

Public Attributes

dnnl_primitive_kind_t primitive_kind
 The kind of primitive. More...
 
dnnl_prop_kind_t prop_kind
 The kind of propagation. More...
 
dnnl_alg_kind_t alg_kind
 The kind of the convolution algorithm. More...
 
dnnl_memory_desc_t src_desc
 Source memory descriptor.
 
dnnl_memory_desc_t diff_src_desc
 Source gradient memory descriptor.
 
dnnl_memory_desc_t weights_desc
 Weights memory descriptor.
 
dnnl_memory_desc_t diff_weights_desc
 Weights gradient memory descriptor.
 
dnnl_memory_desc_t bias_desc
 Bias memory descriptor.
 
dnnl_memory_desc_t diff_bias_desc
 Bias gradient memory descriptor.
 
dnnl_memory_desc_t dst_desc
 Destination memory descriptor.
 
dnnl_memory_desc_t diff_dst_desc
 Destination gradient memory descriptor.
 
dnnl_dims_t strides
 Convolution strides in each spatial dimension.
 
dnnl_dims_t dilates
 Convolution dilates in each spatial dimension.
 
dnnl_dims_t padding [2]
 Padding in each spatial dimension. More...
 
dnnl_data_type_t accum_data_type
 The accumulator data type. Initialized automatically.
 

Detailed Description

A descriptor of a convolution operation.

Examples:
cnn_inference_f32.c, and cpu_cnn_training_f32.c.

Member Data Documentation

◆ primitive_kind

dnnl_primitive_kind_t dnnl_convolution_desc_t::primitive_kind

The kind of primitive.

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

◆ prop_kind

dnnl_prop_kind_t dnnl_convolution_desc_t::prop_kind

◆ alg_kind

dnnl_alg_kind_t dnnl_convolution_desc_t::alg_kind

The kind of the convolution algorithm.

Possible values: dnnl_convolution_direct.

◆ padding

dnnl_dims_t dnnl_convolution_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).


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