Enum Qnn_QuantizationEncoding_t

• Defined in File QnnTypes.h

Enum Documentation

enum Qnn_QuantizationEncoding_t

An enum to specify quantization encoding type structure.

Values:

enumerator QNN_QUANTIZATION_ENCODING_SCALE_OFFSET = 0

Indicates per-tensor scale-offset encoding type. See Qnn_ScaleOffset_t. Support can be checked via QNN_PROPERTY_TENSOR_SUPPORT_QUANTIZATION_ENCODING_SCALE_OFFSET.

enumerator QNN_QUANTIZATION_ENCODING_AXIS_SCALE_OFFSET = 1

Indicates per-axis (e.g. per-channel) scale-offset encoding type. See Qnn_AxisScaleOffset_t. Support can be checked via QNN_PROPERTY_TENSOR_SUPPORT_QUANTIZATION_ENCODING_AXIS_SCALE_OFFSET.

enumerator QNN_QUANTIZATION_ENCODING_BW_SCALE_OFFSET = 2

Indicates bit-width scale-offset encoding type. See Qnn_BwScaleOffset_t. Support can be checked via QNN_PROPERTY_TENSOR_SUPPORT_QUANTIZATION_ENCODING_BW_SCALE_OFFSET.

enumerator QNN_QUANTIZATION_ENCODING_BW_AXIS_SCALE_OFFSET = 3

Indicates bit-width per-axis scale-offset encoding type. See Qnn_BwAxisScaleOffset_t. Support can be checked via QNN_PROPERTY_TENSOR_SUPPORT_QUANTIZATION_ENCODING_BW_AXIS_SCALE_OFFSET.

enumerator QNN_QUANTIZATION_ENCODING_BLOCK = 4

Indicates per-block scale-offset encoding type. See Qnn_BlockScaleOffset_t. Support can be checked via QNN_PROPERTY_TENSOR_SUPPORT_QUANTIZATION_ENCODING_BLOCK_SCALE_OFFSET.

enumerator QNN_QUANTIZATION_ENCODING_BLOCKWISE_EXPANSION = 5

Indicates per-block scale-offset encoding type. See Qnn_BlockScaleOffset_t. Support can be checked via QNN_PROPERTY_TENSOR_SUPPORT_QUANTIZATION_ENCODING_BLOCKWISE_EXPANSION.

enumerator QNN_QUANTIZATION_ENCODING_VECTOR = 6

Indicates VQ compression encoding type. See Qnn_VectorQuantCompression_t. Support can be checked via QNN_PROPERTY_TENSOR_SUPPORT_QUANTIZATION_ENCODING_VQ_COMPRESSION.

enumerator QNN_QUANTIZATION_ENCODING_UNDEFINED = 0x7FFFFFFF