Deep Neural Network Library (DNNL)  1.3.0
Performance library for Deep Learning
Classes | Functions

A primitive to concatenate data by arbitrary dimension. More...

Classes

struct  dnnl::concat
 Tensor concatenation (concat) primitive. More...
 

Functions

dnnl_status_t DNNL_API dnnl_concat_primitive_desc_create (dnnl_primitive_desc_t *concat_primitive_desc, const dnnl_memory_desc_t *dst_desc, int n, int concat_dimension, const dnnl_memory_desc_t *src_descs, const_dnnl_primitive_attr_t attr, dnnl_engine_t engine)
 Creates a primitive descriptor for an out-of-place concatenation primitive. More...
 

Detailed Description

A primitive to concatenate data by arbitrary dimension.

See also
Concat in developer guide

Function Documentation

◆ dnnl_concat_primitive_desc_create()

dnnl_status_t DNNL_API dnnl_concat_primitive_desc_create ( dnnl_primitive_desc_t concat_primitive_desc,
const dnnl_memory_desc_t dst_desc,
int  n,
int  concat_dimension,
const dnnl_memory_desc_t src_descs,
const_dnnl_primitive_attr_t  attr,
dnnl_engine_t  engine 
)

Creates a primitive descriptor for an out-of-place concatenation primitive.

Inputs:

Outputs:

Parameters
concat_primitive_descOutput primitive descriptor.
dst_descDestination memory descriptor.
nNumber of source parameters.
concat_dimensionSource tensors will be concatenated over dimension with this index. Note that order of dimensions does not depend on memory format.
src_descsArray of source memory descriptors with n elements.
attrPrimitive attributes to use (can be NULL).
engineEngine to use.
Returns
dnnl_success on success and a status describing the error otherwise.