Edit

DML_ELEMENT_WISE_IS_INFINITY_OPERATOR_DESC structure (directml.h)

  • Article

Checks each element of InputTensor for IEEE-754 -inf, inf, or both, depending on the given InfinityMode, and places the result (1 for true, 0 for false) into the corresponding element of OutputTensor.

f(x) = isinf(x) && (
       (x > 0 && InfinityMode == DML_IS_INFINITY_MODE_POSITIVE) ||
       (x < 0 && InfinityMode == DML_IS_INFINITY_MODE_NEGATIVE) ||
                 InfinityMode == DML_IS_INFINITY_MODE_EITHER)

Syntax

struct DML_ELEMENT_WISE_IS_INFINITY_OPERATOR_DESC {
  const DML_TENSOR_DESC *InputTensor;
  const DML_TENSOR_DESC *OutputTensor;
  DML_IS_INFINITY_MODE  InfinityMode;
};

Members

InputTensor

Type: const DML_TENSOR_DESC*

The input tensor to read from.

OutputTensor

Type: const DML_TENSOR_DESC*

The output tensor to write the results to.

InfinityMode

Type: DML_IS_INFINITY_MODE

A DML_IS_INFINITY_MODE determining the sign of the infinity to check for.

  • If DML_IS_INFINITY_MODE_EITHER, then 1 will be returned if the element is -inf or inf, otherwise 0.
  • If DML_IS_INFINITY_MODE_POSITIVE, then 1 will be returned if the element is inf, otherwise 0.
  • If DML_IS_INFINITY_MODE_NEGATIVE`, then 1 will be returned if the element is -inf, otherwise 0.

Remarks

Availability

This operator was introduced in DML_FEATURE_LEVEL_2_1.

Tensor constraints

InputTensor and OutputTensor must have the same DimensionCount and Sizes.

Tensor support

DML_FEATURE_LEVEL_3_0 and above

Tensor Kind Supported dimension counts Supported data types
InputTensor Input 1 to 8 FLOAT32, FLOAT16
OutputTensor Output 1 to 8 UINT8

DML_FEATURE_LEVEL_2_1 and above

Tensor Kind Supported dimension counts Supported data types
InputTensor Input 4 FLOAT32, FLOAT16
OutputTensor Output 4 UINT8

Requirements

Requirement Value
Minimum supported client Windows 10, version 2004 (10.0; Build 19041)
Minimum supported server Windows Server, version 2004 (10.0; Build 19041)
Header directml.h