Enum Qnn_Precision_t¶
Defined in File QnnTypes.h
Enum Documentation¶
-
enum Qnn_Precision_t¶
An enum which defines the different precision modes supported by QNN backends. A precision mode may be used to express the math type used in the implementation of an operation.
Values:
-
enumerator QNN_PRECISION_FLOAT32 = 0¶
32-bit Floating point precision. The format of the floating point value is left to backends to choose.
-
enumerator QNN_PRECISION_FLOAT16 = 1¶
16-bit Floating point precision. The format of the floating point value is left to backends to choose.
-
enumerator QNN_PRECISION_UNDEFINED = 0x7FFFFFFF¶
-
enumerator QNN_PRECISION_FLOAT32 = 0¶