Developer Guide and Reference

  • 2022.1
  • 04/11/2022
  • Public Content

struct dnnl_convolution_desc_t

Detailed Documentation

A descriptor of a convolution operation.
dnnl_primitive_kind_t primitive_kind
The kind of primitive.
Used for self-identifying the primitive descriptor. Must be dnnl_convolution.
The kind of propagation.
The kind of the convolution algorithm.
Possible values: dnnl_convolution_direct.
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.
Bias memory descriptor.
dnnl_memory_desc_t diff_bias_desc
Bias gradient memory descriptor.
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.
padding[0] is a padding in the beginning (
), padding[1] is a padding in the end (
dnnl_data_type_t accum_data_type
The accumulator data type. Initialized automatically.

Product and Performance Information


Performance varies by use, configuration and other factors. Learn more at