DML_ELEMENT_WISE_ADD1_OPERATOR_DESC structure (directml.h)

Adds every element in ATensor to its corresponding element in BTensor and places the result into the corresponding element of OutputTensor, with the option for fused activation.

f(a, b) = FusedActivation(a + b)

The fused activation operator description, if provided, then executes the given activation operator on the output.

This operator supports in-place execution, meaning that OutputTensor is permitted to alias one or more of the input tensors during binding.

Syntax

struct DML_ELEMENT_WISE_ADD1_OPERATOR_DESC {
  const DML_TENSOR_DESC   *ATensor;
  const DML_TENSOR_DESC   *BTensor;
  const DML_TENSOR_DESC   *OutputTensor;
  const DML_OPERATOR_DESC *FusedActivation;
};

Members

ATensor

Type: const DML_TENSOR_DESC*

A tensor containing the left-hand side inputs.

BTensor

Type: const DML_TENSOR_DESC*

A tensor containing the right-hand side inputs.

OutputTensor

Type: const DML_TENSOR_DESC*

The output tensor to write the results to.

FusedActivation

Type: _Maybenull_ const DML_OPERATOR_DESC*

An optional fused activation layer to apply after the addition. For more info, see Using fused operators for improved performance.

Fused activation may be used only when the output datatype is FLOAT16 or FLOAT32.

Remarks

Availability

This operator was introduced in DML_FEATURE_LEVEL_2_0.

Tensor constraints

ATensor, BTensor, and OutputTensor must have the same DataType, DimensionCount, and Sizes.

Tensor support

DML_FEATURE_LEVEL_3_0 and above

DML_FEATURE_LEVEL_2_0 and above

Requirements

See also

• Using fused operators for improved performance