Sum

Overview

A primitive to sum multiple tensors. More…

// structs

struct dnnl::sum;

// global functions

dnnl_status_t DNNL_API dnnl_sum_primitive_desc_create(
    dnnl_primitive_desc_t* sum_primitive_desc,
    const dnnl_memory_desc_t* dst_desc,
    int n,
    const float* scales,
    const dnnl_memory_desc_t* src_descs,
    const_dnnl_primitive_attr_t attr,
    dnnl_engine_t engine
    );

Detailed Documentation

A primitive to sum multiple tensors.

See also:

Sum in developer guide

Global Functions

dnnl_status_t DNNL_API dnnl_sum_primitive_desc_create(
    dnnl_primitive_desc_t* sum_primitive_desc,
    const dnnl_memory_desc_t* dst_desc,
    int n,
    const float* scales,
    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) sum primitive.

Parameters:

sum_primitive_desc

Output primitive descriptor.

dst_desc

Destination memory descriptor.

n

Number of source parameters.

scales

Vector of scales to multiply data in each source memory by.

src_descs

Array of source memory descriptors having n elements.

attr

Primitive attributes to use (can be NULL).

engine

Engine to use.

Returns:

dnnl_success on success and a status describing the error otherwise.