DML_DEQUANTIZE_OPERATOR_DESC structure (directml.h)

Article
02/20/2025

TBD

Important

This API is available as part of the DirectML standalone redistributable package (see Microsoft.AI.DirectML version 1.15.0 and later. Also see DirectML version history.

Syntax

struct DML_DEQUANTIZE_OPERATOR_DESC
{
    const DML_TENSOR_DESC* InputTensor;
    DML_QUANTIZATION_TYPE QuantizationType;
    UINT QuantizationTensorCount;
    _Field_size_(QuantizationTensorCount) const DML_TENSOR_DESC* QuantizationTensors;
    const DML_TENSOR_DESC* OutputTensor;
};

Members

InputTensor

Type: const DML_TENSOR_DESC*

The input tensor to read from.

QuantizationType

Type: DML_QUANTIZATION_TYPE

TBD

QuantizationTensorCount

Type: UINT

TBD. This field determines the size of the QuantizationTensors array.

QuantizationTensors

Type: _Field_size_(QuantizationTensorCount) const DML_TENSOR_DESC*

TBD

OutputTensor

Type: const DML_TENSOR_DESC*

The output tensor to write the results to.

Availability

This operator was introduced in DML_FEATURE_LEVEL_6_3.

Tensor constraints

InputTensor, OutputTensor, and QuantizationTensors must have the same DimensionCount.

Tensor support

Tensor	Kind	Supported dimension counts	Supported data types
InputTensor	Input	1 to 8	INT8, INT4, UINT8, UINT4
QuantizationTensors	Array of inputs	1 to 8	FLOAT32, FLOAT16, INT8, INT4, UINT8, UINT4
OutputTensor	Output	1 to 8	FLOAT32, FLOAT16

Share via