DML_TENSOR_DATA_TYPE enumeration (directml.h)

Specifies the data type of the values in a tensor. DirectML operators may not support all data types; see the documentation for each specific operator to find which data types it supports.

Syntax

typedef enum DML_TENSOR_DATA_TYPE {
  DML_TENSOR_DATA_TYPE_UNKNOWN,
  DML_TENSOR_DATA_TYPE_FLOAT32,
  DML_TENSOR_DATA_TYPE_FLOAT16,
  DML_TENSOR_DATA_TYPE_UINT32,
  DML_TENSOR_DATA_TYPE_UINT16,
  DML_TENSOR_DATA_TYPE_UINT8,
  DML_TENSOR_DATA_TYPE_INT32,
  DML_TENSOR_DATA_TYPE_INT16,
  DML_TENSOR_DATA_TYPE_INT8,
  DML_TENSOR_DATA_TYPE_FLOAT64,
  DML_TENSOR_DATA_TYPE_UINT64,
  DML_TENSOR_DATA_TYPE_INT64
} ;

Constants

DML_TENSOR_DATA_TYPE_UNKNOWN Indicates an unknown data type. This value is never valid.
DML_TENSOR_DATA_TYPE_FLOAT32 Indicates a 32-bit floating-point data type.
DML_TENSOR_DATA_TYPE_FLOAT16 Indicates a 16-bit floating-point data type.
DML_TENSOR_DATA_TYPE_UINT32 Indicates a 32-bit unsigned integer data type.
DML_TENSOR_DATA_TYPE_UINT16 Indicates a 16-bit unsigned integer data type.
DML_TENSOR_DATA_TYPE_UINT8 Indicates a 8-bit unsigned integer data type.
DML_TENSOR_DATA_TYPE_INT32 Indicates a 32-bit signed integer data type.
DML_TENSOR_DATA_TYPE_INT16 Indicates a 16-bit signed integer data type.
DML_TENSOR_DATA_TYPE_INT8 Indicates a 8-bit signed integer data type.

Requirements

Requirement	Value
Header	directml.h